Search

Gamma Knife is a medical procedure that is used to treat several types of intracranial disease. The system utilizes gamma rays from Cobalt-60 radiation sources focused at an isocenter and a stereotactic frame system that serves as an immobilization device coordinate system. Treatment is performed by localizing the patient’s disease with a medical imaging study and positioning the diseased area at the focused intersection of the beams. Patient treatment may require multiple treatment positions and varying beam sizes. The treatment position, time, and beam size is determined through a treatment planning process. Traditionally Gamma Knife treatment planning is performed manually by an expert planner. This process can be time consuming and arrival at an optimal plan may depend on the skill of the planner. This work automates the treatment planning process with a multi-module optimization system. First, a kernel regression data mining module compares the treatment volume to a database of past treatment plans to create a set of initial plans. These plans seed a genetic algorithm optimizer that produces an optimized plan. The cost function for the optimization is a weighted average of several traditional metric for assessing stereotactic radiosurgery plan quality. A gradient descent optimizer is utilized to further refine the optimized treatment plan. The developed system was applied to three Gamma Knife planning cases; a solitary metastasis, an acoustic schwannoma, and a pituitary adenoma. The system produced an average percent isodose coverage for the three plans of 94.5% and the average Paddick Conformity index was 0.76 in an average time of 17.16 minutes for the three plans. The system was compared to an expert planner and an optimizer included with the Gamma Knife planning software. The developed system and expert planner performance was essentially equivalent (average percent isodose coverage 95.8%, average Paddick Conformity index 0.70, optimization time 20.52). The developed system performed much better than the Gamma Plan optimizer (average percent isodose coverage 85.8%, average Paddick Conformity index 0.71) however the Gamma Plan optimizer result was obtained quicker (optimization time about 1 minute). The developed system can be utilized for efficient high-quality Gamma Knife treatment planning.

Reconfigurable modular robots can exhibit different specializations by rearranging the same set of parts comprising them. Actuating modular robots can be complicated because of the many degrees of freedom that scale exponentially with the size of the robot. Effectively controlling these robots directly relates to how well they can be used to complete meaningful tasks. This paper discusses an approach for creating provably correct controllers for modular robots from high-level tasks defined with structured English sentences. While this has been demonstrated with simple mobile robots, the problem was enriched by considering the uniqueness of reconfigurable modular robots. These requirements are expressed through traits in the high-level task specification that store information about the geometry and motion types of a robot. Given a high-level problem definition for a modular robot, the approach in this paper deals with generating all lower levels of control needed to solve it. Information about different robot characteristics is stored in a library, and two tools for populating this library have been developed. The first approach is a physics-based simulator and gait creator for manual generation of motion gaits. The second is a genetic algorithm framework that uses traits to evaluate performance under various metrics. Demonstration is done through simulation and with the CKBot hardware platform.

Cellular manufacturing has been an important phenomenon in manufacturing in recent decades. Tremendous amount of work has been done regarding issues such as cell formation, cell loading and job scheduling. However, majority of literature lacks consideration of uncertainty in the problem definition phase, thus methodology. In this dissertation, the impact of uncertainty of demand, processing times and capacity requirements on a cellular manufacturing system (CMS) performance are addressed and stochastic optimization approaches are developed and applied to ten case problems from industrial companies and cellular manufacturing literature. This dissertation consists of mainly three phases, namely: stochastic CMS design, stochastic CMS control and the integrated modeling and analysis of CMS design and CMS control. Capacitated cell formation under the impact of uncertain demand and processing times is defined as the stochastic CMS design problem. On the other hand, cell loading, job sequencing and manpower allocation considering probabilistic demand and processing times are the main issues addressed in the stochastic CMS control phase. Finally, the relationship between stochastic CMS design and stochastic CMS control comprises the "integration" phase. Nonlinear stochastic programming models are developed to optimize each phase and simulation models are also built to validate the results of mathematical optimization and assess manufacturing system performance. To deal with larger problems, as one of the widely used metaheuristic optimization techniques, Genetic Algorithms (GA) is utilized; a GA model is developed and compared with stochastic programming model by using simulation modeling and statistical analysis. Results indicated that stochastic programming can assist with a better decision making on CMS design and control due to its capability of capturing probabilistic nature of problems. In all cases, the proposed stochastic optimization approaches outperformed the conventional deterministic methods. Moreover, the proposed stochastic models let the decision maker to decide the amount of risk to take prior to making design and control related decisions. All in all, I believe that the proposed stochastic optimization-based decision making concepts will open a new corridor in cellular manufacturing research. On the other hand, the proposed approaches can easily be implemented in other popular industrial engineering problem domains including supply chain, healthcare, transportation and logistics.

Deep Learning networks are a new type of neural network that discovers important object features. These networks determine features without supervision, and are adept at learning high level abstractions about their data sets. These networks are useful for a variety of tasks, but are difficult to train. This difficulty is compounded when multiple networks are trained in a layered fashion, which results in increased solution complexity as well as increased training time. This paper examines the use of Genetic Algorithms as a training mechanism for Deep Learning networks, with emphasis on training networks with a large number of layers, each of which is trained independently to reduce the computational burden and increase the overall flexibility of the algorithm. This paper covers the implementation of a multilayer deep learning network using a genetic algorithm, including tuning the genetic algorithm, as well as results of experiments involving data compression and object classification. This paper aims to show that a genetic algorithm can be used to train a non trivial deep learning network in place of existing methodologies for network training, and that the features extracted can be used for a variety of real world computational problems.

A key limitation in accommodating continuing air traffic growth is the fixed airspace structure (sector boundaries), which is largely determined by historical flight profiles that have evolved over time. The sector geometry has stayed relatively constant despite the fact that route structures and demand have changed dramatically over the past decade. Dynamic Airspace Sectorization (DAS) is a concept where the airspace is redesigned dynamically to accommodate changing traffic demands. Various methods have been proposed to dynamically partition the airspace to accommodate traffic growth while also considering other sector constraints and efficiency metrics. However, these approaches suffer several operational drawbacks, and their computational complexity increases exponentially as the airspace size and traffic volume increase. In this thesis, I experimentally evaluate and identify gaps in existing 3D sectorization methods, and propose an improved Agent Based Model (iABM) to address these gaps. I also propose three additional models using KD-Tree, Support Plane Bisection (SPBM) and Constrained Voronoi Diagrams (CVDM) in 3D, to partition the airspace to satisfy the convexity constraint and overcome high computational cost inherent in agent-based approaches. I then look into optimizing the airspace sectors generated by these four models (iABM, KD-Tree, SPBM, and CVDM), using a multi-objective optimisation approach with Air Traffic Controller (ATC) task load balancing, average sector flight time, and minimum distance between sector boundaries and traffic flow crossing points as the three objectives. The performance and efficiency of the proposed models are demonstrated by using sample air traffic data. Experimental results show that all the approaches have strengths and weaknesses. iABM has the best performance on task load balancing, but it can't satisfy the convexity constraint. SPBM and CVDM perform worse than iABM on task load balancing but better on average sector flight time, and they can satisfy the convexity constraint. The KD-tree based model is the most efficient, but not effective as it performed poorly on the given objectives because of its representational bias, which also limits its use in an operational environment. To further investigate SPBM and CVDM for national airspace sectorization, a real time air traffic monitoring and advisory system, called TOP-LAT (Trajectory Optimization and Prediction of Live Air Traffic), is developed and implemented. TOP-LAT is a real time system, synthesizing real time air traffic data to measure and analyse airspace capacity, airspace safety, air traffic flow and aviation emission, to enable ATM participants to access timely, accurate and reliable information for ATM decisions. TOP-LAT provides an ATM environment to evaluate and investigate the advanced ATM concepts, such as DAS. A number of experiments of Australian airspace sectorization by the two proposed DAS models are conducted in this thesis. In these experiments, the current and projected air traffic demands are generated based on public statistics, and some future ATM concepts (e.g. User Preferred Trajectory) are prototyped in order to investigate the performances of the proposed models. The results show that both models have advantages over the current airspace sector configurations in terms of task load balancing, longer flight sector time, larger minimum distance between sector boundaries and traffic flow crossing points, and reduced maximum task load for ATC. These experiments also show that Both models have the capability to be compatible with other advanced ATM concepts. However, no single approach can meet all complex air traffic management objectives. It is the air traffic flow pertaining to the kind of airspace and the associated traffic complexity which can determine the best approach for dynamic sectorization.

Aviation has grown dramatically over the last decades. Worldwide air traffic is expected to continue to grow at rates of 3-5% per year. Air traffic currently experiences long delays at major airports, noticeable impact on the environment, and high work load on air traffic controllers. Future operational concepts to deal with these problems and to support sustainable growth of the Air Navigation Industry have been defined by International Civil Aviation Organization (ICAO) and major Air Traffic Control (ATC) authorities including both Eurocontrol and Federal Aviation Administration (FAA). One of these concepts is User Preferred Routes (UPR), representing the routes with the best business outcome from the airspace user s perspective. Aircraft would generally be free to fly user-preferred routes, and modify their trajectories without, or with minimal, intervention by ATC. There has been limited research on optimisation methods for UPR and on investigating the impact of UPR on the environment, delays, safety, etc. In this thesis a framework is designed and developed to find user preferred routes for flights and evaluate the UPR concept. The main contribution of the thesis is a set of three different methods for finding UPR. Through the experiments with the two first methods, which are based on Genetic Algorithms and Learning Classifier Systems and act as the black and white box optimisation approaches, we found the pros and cons for these different optimisation philosophies. The third method is a scalable algorithm that is fast and reliable. It can optimise routes efficiently for different mixes of users. To enable the investigation and optimisation of UPR, a simulation and evaluation environment is established. In this environment, I developed a real time weather system to retrieve and process weather data. This data is then stored in a database for aviation decision support in general and for aircraft UPR in particular. Second, flight and weather scenarios are designed to test UPR methods and assess user preferred routes provided by these methods. Third, a segment based simulation environment is developed to simulate any type of route segment (climb, cruise, or descent). The simulation uses point mass model of aircraft, and is implemented in a continuous environment, as well as in a fast time mode. This simulation environment is applied effectively and flexibly to evaluate UPR. The impact of different emissions on the environment is measured effectively. I developed a real time flight data management system with algorithms to process, estimate, and integrate information for every flight in the airspace and to construct flight objects. Then another real time system is developed to estimate aviation emission using the flight objects provided by the flight data management system. The aviation emission is stored in 4-D database to analyse the impact of aviation emission on the environment. A number of models for emissions analysis are designed and implemented. The models developed here are then used for the calculation and analysis of fuel and emissions for UPR routes. I demonstrate that, if UPR routes that have been optimised either vertically or horizontally, they can help to reduce on average about 3% and 2% fuel burn in comparison with the original flight plans respectively. If UPR routes are optimised both vertically and horizontally, they can save on average 5% fuel burn. I also demonstrate that the delay in departure times of UPR and non-UPR flights is insignificant (the average delay is less than 1 minute).

Solving Constrained Optimization Problems (COPs) has been an important research topic in the optimization, operation research, and computer science domains. In COPs, an optimal solution must satisfy some feasibility condition which makes them harder when solving as compared to their unconstrained counterpart. In the literature, many Evolutionary Algorithms (EAs) have been reported for solving COPs. In those algorithms, the constraint handling is recognized as a very important component. However, almost all algorithms deal with constraints indirectly. Basically, in every generation, the constraint violation information is used to rank the individuals. Once the selection has been completed for crossover and mutation, those methodologies have no influence on the search process for constraint handling. This may be the reason for the slow convergence and the tendency to become trapped in local optima, when solving COPs using EAs. In this research, a Hybrid Evolutionary Algorithm (HEA), that combines EAs with Constraint Consensus (CC) methods, has been proposed. The CC methods are well-known search approaches that direct infeasible solutions towards the feasible space. Moreover, in the literature, there are many different CC methods with different characteristics. Such methods can rapidly improve the feasibility of individuals. So when combined with EAs, they can help the algorithm convergence faster. This thesis develops the HEAs that use either a single or a mix of three different CC methods with EAs. Two different EAs, namely Differential Evolution (DE) and Genetic Algorithm (GA), are tested with the CC methods. In addition, to show the applicability of the developed algorithm, a real world problem has been solved. HEA uses an adaptive mechanism to decide the appropriate mix of the CC methods in every generation. For further enhancement of the performance of HEA, a local search algorithm has also been included and tested. All of the above algorithms were run to solve a well-known set of 24 benchmark constrained problems. The experimental results show that the inclusion of CC methods with EAs, not only improves the quality of solutions, but also reduces the computational time. Overall, they are thus either better, or competitive, when compared with the state-of-the-art algorithms.

Genetic algorithms are an attractive choice for researchers and practitioners of design optimization. Their advantages are many as compared to classical optimization approaches. However, there are a number of areas to look into so as to make them more practically relevant. This thesis describes original contributions to three such areas Robust Design Optimization, Constraint Handling and Co-Kriging. Robust design optimization using genetic algorithms is computationally expensive because of the high number of function evaluations involved. Hence, it is often simply not possible to use such procedures in practice. We propose some enhancements to an existing robust design optimization algorithm by making it less resource intensive and hence more practically applicable. We study a number of Bi-objective Engineering design optimization problems to understand and demonstrate the benefits of the proposed enhancements. A highlight of this study is the robust design optimization study of an Axisymmetric Scramjet configuration. Next, an interesting and novel constraint handling technique based on the so-called Most probable point (of failure) is described. The proposed method is applicable to optimization problems with equality and active inequality constraints. We present the findings of the proposed approach by considering a test bed of a number of Single and Multi-objective optimization problems. The test bed consists of standard synthetic test problems and engineering design optimization problems. A key contribution here is the formulation of a number of Multi-objective equality constrained test problems. Finally, an interesting modification to Co-Kriging is proposed. Co-Kriging seeks to combine training data obtained at multiple levels of fidelity to enhance the performance of Kriging. It however introduces a number of additional hyper-parameters as compared to Kriging. Determining their optimal values is critical to obtaining a successful Co-Kriging model. A genetic algorithm based optimization technique to find the values of these hyper-paramaters is proposed. The proposed technique also does away with the need to evaluate the likelihood functions and hence dealing with large data sets become more tractable. A number of mathematical examples demonstrate the application of the proposed technique. As a more practical example, data sets obtained from a Scramjet design optimization exercise is considered. The results presented as part of each of the above areas will clearly demonstrate the potential of the proposed techniques in making the use of genetic algorithms in design optimization more practical and viable. The findings also helped us in coming up with a number of avenues for future work in these directions which concludes this thesis.

The goal for this paper and my research is to reduce overall cost associated with electricity use at UC Merced. UC Merced presents itself as a unique opportunity for to model integration and optimization of renewable energy sources. It will be discussed exactly what makes UC Merced unique and how UC Merced has set a path towards higher energy efficiency on a community level. Furthermore, I will discuss difficulties involved with integrating renewable resources and then proceed to analyze techniques for further optimization as UC Merced continues its path towards zero net energy. One of these optimization techniques, genetic algorithms; I will discuss in some detail as it was the technique chose to verify the results of the optimization. The main goal of this study is to determine the effect of moving UC Merced‘s Central Plant load closer to or completely during daylight hours when there is inexpensive (solar) energy available or during the night time when energy pricing is minimum. While it seems logical to shift the cooling load, it has yet to be quantitatively shown that such load shifting would be more cost effective. Genetic algorithm (GA)-based Artificial Neural Network (ANN) models are used for demand and energy production forecasting and then GA based cost optimization is performed to find optimum time window for load shifting. We determined that loading shifting can be beneficial and the associated savings are presented for both summer and winter seasons.

Prognostics and timely maintenance of components are critical to the continuing operation of a system. By implementing prognostics, it is possible for the operator to maintain the system in the right place at the right time. However, the complexity in the real world makes near-zero downtime difficult to achieve partly because of a possible shortage of required service parts. This is realistic and quite important in maintenance practice. To coordinate with a prognostics-based maintenance schedule, the operator must decide when to order service parts and how to compete with other operators who also need the same parts. This research addresses a joint decision-making approach that assists two operators in making proactive maintenance decisions and strategically competing for a service part that both operators rely on for their individual operations. To this end, a maintenance policy involving competition in service part procurement is developed based on the Stackelberg game-theoretic model. Variations of the policy are formulated for three different scenarios and solved via either backward induction or genetic algorithm methods. Unlike the first two scenarios, the possibility for either of the operators being the leader in such competitions is considered in the third scenario. A numerical study on wind turbine operation is provided to demonstrate the use of the joint decision-making approach in maintenance and service part logistics.

Abstract: This thesis deals with finding design points for nonlinear regression models with the possibility that the fitted model is incorrect. The information matrix depends on the parameter in nonlinear situations. We have assumed a range of values of the parameter and have specified a prior on the parameter space. A loss function has been developed and then a minimax approach has been adopted to achieve our goal. We have found an explicit expression for the maximized loss and a numerical minimization of it has been done by a genetic algorithm (GA). The whole approach has been implemented by considering some well-known nonlinear functions. We see that changing the values of the parameter of the prior density have effect on design points. However, changing the tuning constants of GA does not alter the design points noticeably. This indicates that we have obtained the minimizing design.

Abstract: Optimization techniques, in conjunction with a finite element thermal model, are used in this thesis to optimize the temperature profile (i.e. cooling rate and coiling temperature) of a steel skelp during laminar cooling. Optimization parameters include skelp velocity, laminar cooling bank configuration, as well as side-spray conditions. The optimization techniques include two stochastic optimization methods (Genetic Algorithms and Particle swarm optimization) and one deterministic method (The branch-and-bound). A comparison between optimization methods showed that the branch-and-bound method can achieve global optimum faster than the stochastic techniques. The branch-and-bound method was used to set the coiling temperature, using three different cooling strategies (early, late and constant cooling), to reach the specified coiling temperature (550°C). Also the temperature profile optimizations was done, in order to maximize volume of the steel strip, which cool through a desired zone in Continues cooling transformation diagram, was done using branch-and-bound method.

The problem studied in this research is related to cellular manufacturing systems.This research aims to fulfill the gap in manufacturing scheduling literature, in the sense of achieving maximum profit as well as considering multiple due dates of each job and also satisfying the customer order by finding the optimal transportation method in a cellular manufacturing system.In this research, Mathematical Models and Genetic Algorithms are developed and proposed as solution techniques. Moreover, performance of these alternative methods is compared. Finally, it is revealed that the Genetic Algorithms approach is effective in terms of computation time for large problem sizes compared to Mathematical Modeling.

Multi-objectivization is the process of reformulating a single-objective problem into a multi-objective problem and solving it with a multi-objective method in order to provide a solution to the original single-objective problem. Multi-objectivization differs from other traditional divide-and-conquer techniques-the method splits the objective function rather than the search space. Prior to recent evidence, such reformulations were thought to make a problem more complex. However, more recent research suggests that decomposition in the objective space can lead to useful optimization techniques when coupled with a population-based search.Machine based optimization techniques are varied but often based upon an analogy to real-world phenomena. A family and sub-family of algorithms called evolutionary and genetic algorithms respectively are inspired by Darwin's survival-of-the-fittest theory and the concept of modeling evolutionary process. These algorithms manage a population of solutions in a global search process that recombines and creates new solutions in order to generate useful solutions to hard problems. A sub-type of Evolutionary Algorithms (EAs) called Multiple Objective Evolutionary Algorithms (MOEAs) are designed to search for solutions to problems formulated with multiple-objectives. Multiple-objective problems have two or more partially competing objectives such as minimization of cost and maximization of safety. The application of MOEAs to solve problems that in their most natural formulation have a single objective is relatively new. This work investigates Genetic Algorithms (GAs) and their close relatives MOEAs in both a general categorical sense and as they are applied to multi-objectivization. A diversity classification framework for GAs is proposed. Furthermore, multi-objectivization techniques are examined. Through study of an abstract problem, job-shop scheduling problems, and the Traveling Salesman Problem, principles governing the design decisions for multi-objectivization are identified. Two ways in which multi-objectivization creates beneficial search results are theorized: a theory of how multi-objectivization effectively pairs fitness improvements with fitness decrements is developed and a theory for how multi-objectivization circumvents non-linear interactions is proposed. Evidence for both of these beneficial effects is provided through a series of computational experiments. Two prevalent multi-objectivization techniques are compared both analytically and through these experiments. A third, more general version of the studied techniques is proposed with results showing robust performance across a variety of computational budgets.

The one-dimensional bin-packing problem (BPP) is a well known problem in the realm of operations research. In the BPP, objects with pre-defined "weights" are packed into bins, each having a maximum weight capacity. The goal is to minimize the number of bins needed to hold objects. In a Genetic Algorithm for this problem, an individual candidate solution consists of a permutation of object~ representing their order of placement. A heuristic is then used to place objects in bins according to their specified order. The heuristic chosen is generally First Fit, Best Fit, or Worst Fit. The number of bins needed to store the objects is a solution's fitness. In Genetic Algorithms, the fitness value of a solution represents its likelihood to pass on its genetic material to new generations. In the case of the BPP, the less the number of bins needed, the greater a solution's fitness. To create new generations, solutions are "crossed-over" to combine features of their particular solutions and create a new organism. New generations are then created in a cycle until a predetermined limit has been reached and the best organism found becomes the solution to the problem. Many situations arise where two solutions may have the exact same fitness value according to the number of bins. However, these two solutions may not be equivalent in terms of solving the BPP. Solutions which pack objects more tightly in bins are more desirable because they have more potential to be improved in future generations. To distinguish these solutions, we propose a second fitness criterion: the free space in the final bin of a solution. By adding this secondary fitness criterion, the performance of the standard GA algorithm for the BPP can be improved by a significant margin. This has implications for any implementation of bin-packing GAs, as the modification is relatively minor and produces better results; therefore being a cost-effective improvement.

Protocol designs targeted at a specific network scenario or performance metric appear promising on paper, but the complexity and cost of implementing and tuning a routing protocol from scratch presents a major bottleneck in the protocol design process. A unique framework called 'Orchestra` is proposed in the literature to support the testing and development of novel routing designs. The idea of the Orchestra framework is to create generic and reusable routing functional components which can be combined to create unique protocol designs customized for a specific performance metric or network setting. The first contribution of this thesis is the development of a generic, modular, scalable and extensible architecture of the Orchestra framework. Once the architecture and implementation of the framework is completed, the second contribution of this thesis is the development of functional components and strategies to design and implement routing protocols for delay tolerant networks (DTNs). DTNs are a special type of ad hoc network characterized by intermittent connectivity, long propagation delays and high loss rate. Thus, traditional ad hoc routing approaches cannot be used in DTNs, and special features must be developed for the Orchestra framework to support the design of DTN routing protocols. The component-based architecture of Orchestra can capture a variety of modules that can be used to assemble a routing protocol. However, manually assembling these components may result in suboptimal designs, because it is difficult to determine what the best combination is for a particular set of performance objectives and network characteristics. The third contribution of the thesis addresses this problem. A genetic algorithm based approach to automate the process of routing protocol design is developed and its performance is evaluated in the context of the Orchestra framework.

Abstract: Multiple-input, multiple-output (MIMO) techniques have been proposed to meet the needs for higher data rates and lower delays in future wireless communication systems. The downlink capacity of multiuser MIMO systems is achieved when the system transmits to several users simultaneously. Frequently, many more users request service than the transmitter can simultaneously support. Thus, the transmitter requires a scheduling algorithm for the users, which must balance the goals of increasing throughput, reducing multiuser interference, lowering delays, ensuring fairness and quality of service (QoS), etc. In this thesis, we investigate the application of genetic algorithms (GAs) to perform scheduling in multiuser MIMO systems. GAs are a fast, suboptimal, low-complexity method of solving optimization problems, such as the maximization of a scheduling metric, and can handle arbitrary functions and QoS constraints. We first examine a system that transmits using capacity-achieving dirty paper coding (DPC). Our proposed GA structure both selects users and determines their encoding order for DPC, which affects the rates they receive. Our GA can also schedule users independently on different carriers of a multi-carrier system. We demonstrate that the GA performance is close to that of an optimal exhaustive search, but at a greatly reduced complexity. We further show that the GA convergence time can be significantly reduced by tuning the values of its parameters. While DPC is capacity-achieving, it is also very complex. Thus, we also investigate GA scheduling with two linear precoding schemes, block diagonalization and successive zero-forcing. We compare the complexity and performance of the GA with "greedy" scheduling algorithms, and find the GA is more complex, but performs better at higher signal-to-noise ratios (SNRs) and smaller user pool sizes. Both algorithms are near-optimal, yet much less complex than an exhaustive search. We also propose hybrid greedy-genetic algorithms to gain benefits from both types of algorithms. Lastly, we propose an improved method of optimizing the transmit covariance matrices for successive zero-forcing. Our algorithm significantly improves upon the performance of the existing method at medium to high SNRs, and, unlike the existing method, can maximize a weighted sum rate, which is important for fairness and QoS considerations.

This thesis examines the performance of genetic algorithm (GA) crossover techniques within two problems: n-queens with poison (NQWP) and processor scheduling (PS). Each problem was analyzed at sizes of 32, 64, and 128, referring to number of queens to be placed and number of single-time-unit processes to be scheduled, respectively. The specific crossover techniques studied were cycle crossover, order crossover, partially mapped crossover, merging crossover, and one-point, two-point, and uniform signature representation crossover, in addition to various greedy approaches. In conjunction with tests that vary crossover techniques, experimentation was performed to determine what percentage of problem constraints (poisoned squares for NQWP or precedence relationships between tasks for PS) makes the problems most difficult to solve, that is, the constraint densities at which optimal solutions require the highest number of GA fitness evaluations. While minor fluctuations in difficulty occur upon variations in fitness function and problem size, the NQWP problem is most difficult around a constraint density of 0.8 and the PS problem is most difficult around constraint densities of 0.2 to 0.3. Even within an individual problem, one crossover technique does not irreproachably outperform others. However, cycle crossover stands out in its performance in the PS problem while merging crossover and uniform signature crossovers most often perform well for NQWP.

As reconfigurable devices' capacities and the complexity of applications that use them increase, the need for self-reliance of deployed systems becomes increasingly prominent. A Sustainable Modular Adaptive Redundancy Technique (SMART) composed of a dual-layered organic system is proposed, analyzed, implemented, and experimentally evaluated. SMART relies upon a variety of self-regulating properties to control availability, energy consumption, and area used, in dynamically-changing environments that require high degree of adaptation. The hardware layer is implemented on a Xilinx Virtex-4 Field Programmable Gate Array (FPGA) to provide self-repair using a novel approach called a Reconfigurable Adaptive Redundancy System (RARS). The software layer supervises the organic activities within the FPGA and extends the self-healing capabilities through application-independent, intrinsic, evolutionary repair techniques to leverage the benefits of dynamic Partial Reconfiguration (PR). A SMART prototype is evaluated using a Sobel edge detection application. This prototype is shown to provide sustainability for stressful occurrences of transient and permanent fault injection procedures while still reducing energy consumption and area requirements. An Organic Genetic Algorithm (OGA) technique is shown capable of consistently repairing hard faults while maintaining correct edge detector outputs, by exploiting spatial redundancy in the reconfigurable hardware. A Monte Carlo driven Continuous Markov Time Chains (CTMC) simulation is conducted to compare SMART's availability to industry-standard Triple Modular Technique (TMR) techniques. Based on nine use cases, parameterized with realistic fault and repair rates acquired from publically available sources, the results indicate that availability is significantly enhanced by the adoption of fast repair techniques targeting aging-related hard-faults. Under harsh environments, SMART is shown to improve system availability from 36.02% with lengthy repair techniques to 98.84% with fast ones. This value increases to "five nines" (99.9998%) under relatively more favorable conditions. Lastly, SMART is compared to twenty eight standard TMR benchmarks that are generated by the widely-accepted BL-TMR tools. Results show that in seven out of nine use cases, SMART is the recommended technique, with power savings ranging from 22% to 29%, and area savings ranging from 17% to 24%, while still maintaining the same level of availability.

A Sustainable Autonomic Architecture for Organically Reconfigurable Computing System based on SRAM Field Programmable Gate Arrays (FPGAs) is proposed, modeled analytically, simulated, prototyped, and measured. Low-level organic elements are analyzed and designed to achieve novel self-monitoring, self-diagnosis, and self-repair organic properties. The prototype of a 2-D spatial gradient Sobel video edge-detection organic system use-case developed on a XC4VSX35 Xilinx Virtex-4 Video Starter Kit is presented. Experimental results demonstrate the applicability of the proposed architecture and provide the infrastructure to quantify the performance and overcome fault-handling limitations. Dynamic online autonomous functionality restoration after a malfunction or functionality shift due to changing requirements is achieved at a fine granularity by exploiting dynamic Partial Reconfiguration (PR) techniques. A Genetic Algorithm (GA)-based hardware/software platform for intrinsic evolvable hardware is designed and evaluated for digital circuit repair using a variety of well-accepted benchmarks. Dynamic bitstream compilation for enhanced mutation and crossover operators is achieved by directly manipulating the bitstream using a layered toolset. Experimental results on the edge-detector organic system prototype have shown complete organic online refurbishment after a hard fault. In contrast to previous toolsets requiring many milliseconds or seconds, an average of 0.47 microseconds is required to perform the genetic mutation, 4.2 microseconds to perform the single point conventional crossover, 3.1 microseconds to perform Partial Match Crossover (PMX) as well as Order Crossover (OX), 2.8 microseconds to perform Cycle Crossover (CX), and 1.1 milliseconds for one input pattern intrinsic evaluation. These represent a performance advantage of three orders of magnitude over the JBITS software framework and more than seven orders of magnitude over the Xilinx design flow. Combinatorial Group Testing (CGT) technique was combined with the conventional GA in what is called CGT-pruned GA to reduce repair time and increase system availability. Results have shown up to 37.6% convergence advantage using the pruned technique. Lastly, a quantitative stochastic sustainability model for reparable systems is formulated to evaluate the Sustainability of FPGA-based reparable systems. This model computes at design-time the resources required for refurbishment to meet mission availability and lifetime requirements in a given fault-susceptible missions. By applying this model to MCNC benchmark circuits and the Sobel Edge-Detector in a realistic space mission use-case on Xilinx Virtex-4 FPGA, we demonstrate a comprehensive model encompassing the inter-relationships between system sustainability and fault rates, utilized, and redundant hardware resources, repair policy parameters and decaying reparability.

Optimization is considered an essential step in reinforcing the efficiency of performance and economic feasibility of construction projects. In the past few decades, evolutionary algorithms (EAs) have been widely utilized to solve various types of construction-related optimization problems due to their efficiency in finding good solutions in relatively short time periods. However, in many cases, these existing evolutionary algorithms failed to identify the optimal solution to several optimization problems. As such, it is deemed necessary to develop new approaches in order to help identify better-quality solutions. This doctoral research presents the development of a new evolutionary algorithm, named “Electimize,” that is based on the simulation of the flow of electric current in the branches of an electric circuit. The main motive in this research is to provide the construction industry with a robust optimization tool that overcomes some of the shortcomings of existing EAs. In solving optimization problems using Electimize, a number of wires (solution strings) composed of a number of segments are fabricated randomly. Each segment corresponds to a decision variable in the objective function. The wires are virtually connected in parallel to a source of an electricity to represent an electric circuit. The electric current passing through each wire is calculated by substituting the values of the segments in the objective function. The quality of the wire is based on its global resistance, which is calculated using Ohm’s law. iv he main objectives of this research are to 1) develop an optimization methodology that is capable of evaluating the quality of decision variable values in the solution string independently; 2) devise internal optimization mechanisms that would enable the algorithm to extensively search the solution space and avoid its convergence toward local optima; and 3) provide the construction industry with a reliable optimization tool that is capable of solving different classes of NP-hard optimization problems. First, internal processes are designed, modeled, and tested to enable the individual assessment of the quality of each decision variable value available in the solution space. The main principle in assessing the quality of each decision variable value individually is to use the segment resistance (local resistance) as an indicator of the quality. This is accomplished by conducting a sensitivity analysis to record the change in the resistance of a control wire, when a certain decision variable value is substituted into the corresponding segment of the control wire. The calculated local resistances of all segments of a wire are then normalized to ensure that their summation is equal to the global wire resistance and no violation is made of Kirchhoff’s rule. A benchmark NP-hard cash flow management problem from the literature is attempted to test and validate the performance of the developed approach. Not only was Electimize able to identify the optimal solution for the problem, but also it identified ten alternative optimal solutions, outperforming the existing algorithms. Second, the internal processes for the sensitivity analysis are designed to allow for extensive search of the solution space through the generation of new v wires. Every time a decision variable value is substituted in the control wire to assess its quality, a new wire that might have a better quality is generated. To further test the capabilities of Electimize in searching the solution space, Electimize was applied to a multimodal 9-city travelling salesman problem (TSP) that had been previously designed and solved mathematically. The problem has 27 alternative optimal solutions. Electimize succeeded to identify 21 of the 27 alternative optimal solutions in a limited time period. Moreover, Electimize was applied to a 16-city benchmark TSP (Ulysses16) and was able to identify the optimal tour and its alternative. Further, additional parameters are incorporated to 1) allow for the extensive search of the solution space, 2) prevent the convergence towards local optima, and 3) increase the rate of convergence towards the global optima. These parameters are classified into two categories: 1) resistance related parameters, and 2) solution exploration parameters. The resistance related parameters are: a) the conductor resistivity, b) its cross-sectional area, and c) the length of each segment. The main role of this set of parameters is to provide the algorithm with additional gauging parameters to help guide it towards the global optima. The solution exploration parameters included a) the heat factor, and b) the criterion of selecting the control wire. The main role of this set of parameters is to allow for an extensive search of the solution space in order to facilitate the identification all the available alternative optimal solutions; prevent the premature convergence towards local optima; and increase the rate of convergence towards the global optima. Two TSP instances (Bayg29 and ATT48) are attempted and vi the results obtained illustrate that Electimize outperforms other EAs with respect to the quality of solutions obtained. Third, to test the capabilities of Electimize as a reliable optimization tool in construction optimization problems, three benchmark NP-hard construction optimization problems are attempted. The first problem is the cash flow management problem, as mentioned earlier. The second problem is the time cost tradeoff problem (TCTP) and is used as an example of static optimization. The third problem is a site layout planning problem (SLPP), and represents dynamic optimization. When Electimize was applied to the TCTP, it succeeded to identify the optimal solution of the problem in a single iteration using thirty solution strings, compared to hundreds of iterations and solution strings that were used by EAs to solve the same problem. Electimize was also successful in solving the SLPP and outperformed the existing algorithm used to solve the problem by identifying a better optimal solution. The main contributions of this research are 1) developing a new approach and algorithm for optimization based on the simulation of the phenomenon of electrical conduction, 2) devising processes that enable assessing the quality of decision variable values independently, 3) formulating methodologies that allow for the extensive search of the solution space and identification of alternative optimal solutions, and 4) providing a robust optimization tool for decision makers and construction planners.

With the ongoing technology battles and price wars in today's competitive economy, every company is looking for an advantage over its peers. A particular choice of facility layout can have a significant impact on the ability of a company to maintain lower operational expenses under uncertain economic conditions. It is known that systems with less congestion have lower operational costs. Traditionally, manufacturing facility layout problem methods aim at minimizing the total distance traveled, the material handling cost, or the time in the system (based on distance traveled at a specific speed). The proposed methodology solves the looped layout design problem for a looped layout manufacturing facility with a looped conveyor material handling system with shortcuts using a system performance metric, i.e. the work in process (WIP) on the conveyor and at the input stations to the conveyor, as a factor in the minimizing function for the facility layout optimization problem which is solved heuristically using a permutation genetic algorithm. The proposed methodology also presents the case for determining the shortcut locations across the conveyor simultaneously (while determining the layout of the stations around the loop) versus the traditional method which determines the shortcuts sequentially (after the layout of the stations has been determined). The proposed methodology also presents an analytical estimate for the work in process at the input stations to the closed looped conveyor. It is contended that the proposed methodology (using the WIP as a factor in the minimizing function for the facility layout while simultaneously solving for the shortcuts) will yield a facility layout which is less congested than a facility layout generated by the traditional methods (using the total distance traveled as a factor of the minimizing function for the facility layout while sequentially solving for the shortcuts). The proposed methodology is tested on a virtual 300mm Semiconductor Wafer Fabrication Facility with a looped conveyor material handling system with shortcuts. The results show that the facility layouts generated by the proposed methodology have significantly less congestion than facility layouts generated by traditional methods. The validation of the developed analytical estimate of the work in process at the input stations reveals that the proposed methodology works extremely well for systems with Markovian Arrival Processes.

The fields of population genetics, evolutionary computing, and artificial life have existed as separate domains for a number of decades. Here, I present an ambitious model of the relationship between DNA sequence evolution and the dynamics of sexual reproduction using concepts from these different fields. While models of DNA sequence evolution have been created using continuous-time Markov models, these models do not allow close observation of the processes of mutation, gametogenesis, and sexual reproduction; most of these models do not account for the context that base substitutions have within functional regions of a genome (e.g. coding and untranslated exons, introns, intergenic regions, isochors, promoters, etc). The Genome Evolution Model (GEM) described in this thesis models DNA sequence change in the context of fitness and selective pressure. I describe the operation of GEM, and provide an example of its application to modeling population dynamics. The included experiment, intended to demonstrate how GEM operates, also shows intriguing results suggesting that GEM may be able to model inbreeding depression as a result of decreased genetic diversity.

Learning classifier systems (LCS) have been successful in generating rules for solving classification problems in data mining. The rules are of the form IF condition THEN action. The condition encodes the features of the input space and the action encodes the class label. What is lacking in those systems is the ability to express each feature using a function that is appropriate for that feature. The genetic algorithm is capable of doing this but cannot because only one type of membership function is provided. Thus, the genetic algorithm learns only the shape and placement of the membership function, and in some cases, the number of partitions generated by this function. The research conducted in this study employs a learning classifier system to generate the rules for solving classification problems, but also incorporates multiple types of membership functions, allowing the genetic algorithm to choose an appropriate one for each feature of the input space and determine the number of partitions generated by each function. In addition, three membership functions were introduced. This paper describes the framework and implementation of this modified learning classifier system (M-LCS). Using the M-LCS model, classifiers were simulated for two benchmark classification problems and two additional real-world problems. The results of these four simulations indicate that the M-LCS model provides an alternative approach to designing a learning classifier system. The following contributions are made to the field of computing: 1) a framework for developing a learning classifier system that employs multiple types of membership functions, 2) a model, M-LCS, that was developed from the framework, and 3) the addition of three membership functions that have not been used in the design of learning classifier systems.

Abstract: Digital filters having sharp transition band play a vital role in modern digital signal processing (DSP) applications. Emerging technologies require digital filters to be both computationally efficient in software/hardware realizations. This thesis is concerned with the design and structural-level optimization of sharp transition band digital filters employing the well known frequency response masking (FRM) approach. Unlike the conventional finite impulse response (FIR) based FRM approach, the FRM technique used in this thesis incorporates infinite impulse response (IIR) digital interpolation subfilters, thereby reducing the overall filter order that results in a reduction of hardware complexity. Two realization methods are discussed in this thesis, namely, the bilinear-lossless-discrete-integrators (bilinear-LDI) digital filter design technique, and the lattice wave digital filter (lattice WDF) digital filter design technique. Diversity controlled (DC) genetic algorithm (GA) is employed to optimize both types of IIR based FRM digital filters over the efficient canonical signed digit (CSD) multiplier coefficient space. DCGAs represent FRM digital filters by a binary chromosome and proceed from a population pool of candidate chromosomes to future generations in order to arrive at the desired FRM digital filter satisfying the design specifications. A novel cost-function is used that allows the DCGA to simultaneously optimize both the amplitude-frequency and group-delay frequency response. A fast convergence speed has been observed.

In this study, an improved genetic algorithm (GA) is presented to solve the multidimensional 0-1 knapsack problem (MKP). The MKP is a well-known combinatorial optimization problem and has received wide attention from the operations research community for decades. Although recent advances in computing and optimization technologies have made the solution of small and medium size instances possible, this NP-hard problem, in general, still remains one of the challenging problems yet to be solved.Of the various algorithms developed to solve the MKP, GA seems to be one of the best methods pointed out in the literature. A GA is an iterative search procedure that simulates the evolution process of a population of individuals based on natural selection and genetics. A GA typically starts with a random initial population and uses genetic operators such as crossover and mutation to yield new offspring to replace individuals of current population. GAs, though have been successful in solving MKPs, could be slow in converging to an optimal or near optimal solution.An improved GA is proposed in this study that aims at exploring the use of greedy heuristics and methods to generate multiple diverse solutions to speed GA convergence. Path re-linking (PR), a method to generate new solutions by exploring trajectories that connect high quality solutions, is used to combine elite solutions to further improve the quality of solutions. The combination of uniform crossover and PR allows the integration of randomization and elite solutions analysis to achieve a balance of intensification and diversification to further improve the quality of solutions.Computational studies of benchmark problems suggest that the proposed algorithm was able to quickly achieve good solutions while avoiding being trapped in premature convergence and is on par with some of the state-of-the-art algorithms in the literature. This study demonstrates a systematic method to explore heuristics to generate population generation with diversity, which could significantly influence the convergence of a GA to best solutions. Nevertheless, as our computational results suggest, randomization in crossover is critical for a GA in its overall performance to achieve better quality solutions.

Artificial neural networks (ANNs) are an abstraction of the low-level architecture of biological brains that are often applied in general problem solving and function approximation. Neuroevolution (NE), i.e. the evolution of ANNs, has proven effective at solving problems in a variety of domains. Information from the domain is input to the ANN, which outputs its desired actions. This dissertation presents a new NE algorithm called Hypercube-based NeuroEvolution of Augmenting Topologies (HyperNEAT), based on a novel indirect encoding of ANNs. The key insight in HyperNEAT is to make the algorithm aware of the geometry in which the ANNs are embedded and thereby exploit such domain geometry to evolve ANNs more effectively. The dissertation focuses on applying HyperNEAT to tactical and strategic decision domains. These domains involve simultaneously considering short-term tactics while also balancing long-term strategies. Board games such as checkers and Go are canonical examples of such domains; however, they also include real-time strategy games and military scenarios. The dissertation details three proposed extensions to HyperNEAT designed to work in tactical and strategic decision domains. The first is an action selector ANN architecture that allows the ANN to indicate its judgements on every possible action all at once. The second technique is called substrate extrapolation. It allows learning basic concepts at a low resolution, and then increasing the resolution to learn more advanced concepts. The iii final extension is geometric game-tree pruning, whereby HyperNEAT can endow the ANN the ability to focus on specific areas of a domain (such as a checkers board) that deserve more inspection. The culminating contribution is to demonstrate the ability of HyperNEAT with these extensions to play Go, a most challenging game for artificial intelligence, by combining HyperNEAT with UCT

Social scientists and other users of large data sets often desire a model to predict the probability that some condition exists, such as the probability that a person has diabetes or that a credit card transaction will be fraudulent. In general, this can be done by data mining techniques, which allow multiple records of data composed of numerous independent variables and one dependent variable to be examined in a statistical fashion to make a predictive model. One particular technique used is logistic regression. Logistic regression forms a predictive model based on a set of independent variables by assigning coefficients to these variables to maximize a non-linear function. The state-of-the-art for creating logistic regression models requires the modeler to select independent variables and the use of an iterative search technique to solve the underlying non-linear optimization. This dissertation investigates the use of genetic algorithms for creating a logistic regression model. The use of this optimization technique facilitates resolution of two critiques of the state-of-the-art: (1) user selection of independent variables allows bias from the user to enter into the logistic regression model; (2) the iterative optimization method used can result in sub-optimal models being accepted. The use of genetic algorithms in the place of the current optimization technique effectively addresses these concerns. Data of increasing complexity are considered, from one to several independent variables. In response, genetic algorithms that allow for increasing flexibility are developed. Through extensive computational studies, a robust genetic algorithm that allows for selection of independent variables and setting of parameter values is developed. The power of the developed approach is demonstrated in a case study of general aviation accident data with five hundred cases and thirteen independent variables.

The objective of this thesis is to provide an efficient and accurate corrective solution to a system that is on verge of voltage collapse. This thesis describes, in detail, the development of an optimization scheme that aims to alleviate power system instability and voltage collapse condition based on the principles of an evolutionary approach called Genetic Algorithm. The state of a system is determined using a voltage stability identifier termed Collapse Proximity Index (CPI) and the critical loading condition is identified. Applying principles of Genetic Algorithm, the critical system is brought back to a stable operating region. The sequential procedure and application of this scheme is primarily discussed in this thesis. The thesis is structured to include theoretical discussion of the Collapse Proximity Index, development of a Genetic Algorithm - based solution to the voltage collapse problem and its simulated implementation on a test system, along with result analysis and suggestions for future development. Conclusions are drawn based on the efficiency of the application in maintaining system stability.

Abstract: Regarding the problems of resource allocation in OFDMA-based wireless communication systems, much of the research effort mainly focuses on finding efficient power control and subcarrier assignment policies. With systems employing multicast transmission, the available schemes in literature are not always applicable. Moreover, the existing approaches are particularly inaccessible in practical systems in which there are a large number of OFDM subcarriers being utilized, as the required computational burden is prohibitively high. The ultimate goal of this research is therefore to propose affordable mechanisms to flexibly and effectively share out the available resources in multicast wireless systems deploying OFDMA technology. Specifically, we study the resource distribution problems in both conventional and cognitive radio network settings, formulating the design problems as mathematical optimization programs, and then offering the solution methods. Suboptimal and optimal schemes with high performance and yet of acceptable complexity are devised through the application of various mathematical optimization tools such as genetic algorithm and Lagrangian dual optimization. The novelties of the proposed approaches are confirmed, and their performances are verified by computer simulation with the presentation of numerical examples to support the findings.

The main characteristic of the problem is that it considers a synchronized environment. A scheduling problem is handled in a manufacturing system that is using uniform time bucket approach for synchronization. There are three main aspects of the problem. The first aspect is the allocation of manpower. The manpower levels are determined in order to maximize production rate. Then, cell loading issue is addressed for multiple cells. Finally, a schedule is generated for each cell in order to minimize the number of periods where available manpower is exceeded. Two methods are used in order to solve these three main aspects of the problem, namely mathematical modeling and genetic algorithm. It has been concluded that mathematical modeling is better for cell loading and genetic algorithm is better for product sequencing. Also, minimizing similarity among products worked better during cell loading phase. Finally, sequential solution of cell loading and product sequencing was better compared to simultaneous solution of these two phases.

Neural networks with chaotic baseline behavior are interesting for their experimental bases in both biological relevancy and engineering applicability. In the engineering case, the literature still lacks a robust study of the interrelationship between particular chaotic baseline network dynamics and 'online' or 'driving' inputs. We ask the question, for a particular neural network with chaotic baseline behavior, what periodic inputs of minimal magnitude have a stabilizing effect on network dynamics? A genetic algorithm is developed for the task. A systematic comparison of different genetic operators is carried out where each operator-combination is ranked by the optimality of solutions found. The algorithm reaches acceptable results and _finds input sequences with largest elements on the order of 10^3. Lastly, an illustration of the complexity of the fitness space is produced by brute-force sampling period-2 inputs and plotting a fitness map of their stabilizing effect on the network.

Most real world optimization problems, and their corresponding models, are complex. This complexity arises from different sources, such as existence of the constraints, function characteristics, and high dimensionality. Evolutionary Algorithms (EAs) and specially Genetic Algorithms (GAs) have proven themselves as efficient optimization techniques over the last two decades; as they have the ability to overcome the drawbacks of conventional optimization methods. Therefore, this thesis addresses the GAs as a solution methodology for solving Constrained Optimization Problems (COPs). In COPs for practical applications, it is more likely that the optimal solution lies on the feasible region boundary. Utilizing this feature, this thesis introduces a new genetic algorithm for solving small-scale COPs. A new ranking and selection scheme is introduced in conjunction with both a new crossover method based on three parents, and a mixed mutation between two currently existing mutation methodologies. A number of well known benchmark problems have been solved and compared with the state of the art algorithms, and the proposed algorithm shows a competitive and even superior performance for some problems. In addition, a detailed parametric analysis is provided to show the individual effect of each of the proposed components. Furthermore, this thesis introduces another algorithm that breaks down the complexity of the constrained optimization process into smaller dimensions. Every sub-component of the algorithm maintains a part of the problem, and the whole problem optimization is treated through a designed communication process. This algorithm deals with a special problem structure, in which the problems are entirely or almost decomposable based on what is called in the literature as a block angular structure. The proposed algorithm decomposes the constraints as well as the chromosomes. It facilitates solving such problems both with and without overlapping variables between the sub-components. Some experiments have been carried out to show how the designed communication process controls the optimization and what the best parameter settings are. Then, the algorithm has been implemented in a parallel environment on a scalable practical test problem and using this shows how the proposed algorithm outperforms the other single population based algorithms in higher dimensions.

Many real-world decision processes require solving optimization problems which may involve different types of constraints such as inequality and equality constraints. The hurdles in solving these Constrained Optimization Problems (COPs) arise from the challenge of searching a huge variable space in order to locate feasible points with acceptable solution quality. Over the last decades Evolutionary Algorithms (EAs) have brought a tremendous advancement in the area of computer science and optimization with their ability to solve various problems. However, EAs have inherent difficulty in dealing with constraints when solving COPs. This thesis presents a new Agent-based Memetic Algorithm (AMA) for solving COPs, where the agents have the ability to independently select a suitable Life Span Learning Process (LSLP) from a set of LSLPs. Each agent represents a candidate solution of the optimization problem and tries to improve its solution through cooperation with other agents. Evolutionary operators consist of only crossover and one of the self-adaptively selected LSLPs. The performance of the proposed algorithm is tested on benchmark problems, and the experimental results show convincing performance. The quality of individuals in the initial population influences the performance of evolutionary algorithms, especially when the feasible region of the constrained optimization problems is very tiny in comparison to the entire search space. This thesis proposes a method that improves the quality of randomly generated initial solutions by sacrificing very little in diversity of the population. The proposed Search Space Reduction Technique (SSRT) is tested using five different existing EAs, including AMA, by solving a number of state-of-the-art test problems and a real world case problem. The experimental results show SSRT improves the solution quality, and speeds up the performance of the algorithms. The handling of equality constraints has long been a difficult issue for evolutionary optimization methods, although several methods are available in the literature for handling functional constraints. In any optimization problems with equality constraints, to satisfy the condition of feasibility and optimality the solution points must lie on each and every equality constraint. This reduces the size of the feasible space and makes it difficult for EAs to locate feasible and optimal solutions. A new Equality Constraint Handling Technique (ECHT) is presented in this thesis, to enhance the performance of AMA in solving constrained optimization problems with equality constraints. The basic concept is to reach a point on the equality constraint from its current position by the selected individual solution and then explore on the constraint landscape. The technique is used as an agent learning process in AMA. The experimental results confirm the improved performance of the proposed algorithm. This thesis also proposes a Modified Genetic Algorithm (MGA) for solving COPs with equality constraints. After achieving inspiring performance in AMA when dealing with equality constraints, the new technique is used in the design of MGA. The experimental results show that the proposed algorithm overcomes the limitations of GA in solving COPs with equality constraints, and provides good quality solutions.

Building an intelligent agent model from scratch is a difficult task. Thus, it would be preferable to have an automated process perform this task. There have been many manual and automatic techniques, however, each of these has various issues with obtaining, organizing, or making use of the data. Additionally, it can be difficult to get perfect data or, once the data is obtained, impractical to get a human subject to explain why some action was performed. Because of these problems, machine learning from observation emerged to produce agent models based on observational data. Learning from observation uses unobtrusive and purely observable information to construct an agent that behaves similarly to the observed human. Typically, an observational system builds an agent only based on prerecorded observations. This type of system works well with respect to agent creation, but lacks the ability to be trained and updated on-line. To overcome these deficiencies, the proposed system works by adding an augmented force-feedback system of training that senses the agents intentions haptically. Furthermore, because not all possible situations can be observed or directly trained, a third stage of learning from practice is added for the agent to gain additional knowledge for a particular mission. These stages of learning mimic the natural way a human might learn a task by first watching the task being performed, then being coached to improve, and finally practicing to self improve. The hypothesis is that a system that is initially trained using human recorded data (Observational), then tuned and adjusted using force-feedback (Instructional), and then allowed to perform the task in different situations (Experiential) will be better than any individual step or combination of steps.

OCEON-P is a computer program whose purpose is to minimize the levelized fuel cycle cost over a multi-cycle planning horizon. It integrates a core simulator, fuel cycle cost calculator and mathematical optimization engine. The accuracy of the predicted fuel cycle cost, whose minimization guides the optimization of the decision variables, is directly related to the fidelity of the reactor core simulator used by the program. Unfortunately, high fidelity core simulators also require longer run times. To improve these run times, this project sought to parallelize the optimization process so that multiple processors may share the computational burden. In addition, an effort was made to reduce the number of fuel cycles that must be examined to complete the optimization, which also reduces the computer run times. Parallelization of the process was introduced by the replacement of the current serial simulated annealing method with parallel simulated and genetic algorithms. It was hoped that genetic algorithms would also reduce the number of fuel cycles that must be examined during the optimization search. However, it was found that although genetic algorithms could find the same caliber of best solutions as simulated annealing, simulated annealing could produce a better family of acceptable solutions. Furthermore, parallel simulated annealing was able to reproduce the same quality and robustness of serial simulated annealing while decreasing run times significantly through use of multiple processors.

This thesis presents an implementation of a learning classifier system which learns good database schema. The system is implemented in Java using the NetBeans development environment, which provides a good control for the GUI components. The system contains four components: a user interface, a rule and message system, an apportionment of credit system, and genetic algorithms. The input of the system is a set of simple database schemas and the objective for the classifier system is to keep the good database schemas which are represented by classifiers. The learning classifier system is given some basic knowledge about database concepts or rules. The result showed that the system could decrease the bad schemas and keep the good ones.

Lossy image compression algorithms sacrifice perfect imagereconstruction in favor of decreased storage requirements. Modernlossy compression schemes, such as JPEG2000, rely upon the discretewavelet transform (DWT) to achieve high levels of compression whileminimizing the loss of information for image reconstruction. Somecompression applications require higher levels of compression thanthose achieved through application of the DWT and entropy coding. Insuch lossy systems, quantization provides high compression rates atthe cost of increased distortion. Unfortunately, as the amount ofquantization increases, the performance of the DWT for accurateimage reconstruction deteriorates. Previous research demonstratesthat a genetic algorithm can improve image reconstruction in thepresence of quantization error by replacing the wavelet filtercoefficients with a set of evolved coefficients.This dissertation develops a methodology for the evolution ofdigital filters capable of outperforming the DWT for imagereconstruction at a given compression rate in the presence ofquantization error. This dissertation compares potential fitnessmeasures for evaluating reconstruction error. Experiments comparethe usefulness of local versus standard population initializationand mutation operators. In order to perform an efficient yetthorough traversal of the search space, several recombinationoperators developed specifically for real-valued evolution areevaluated. Additionally, this dissertation presents and develops anovel technique to emphasize the reconstruction of the high-spacialfrequency areas of an image through use of edge detection algorithmsand focused evolution. An analysis of the ease of traversal throughthe fitness landscapes defined by various image quality measuressupports the development of a framework for evolving robust imagetransform filters.Particular emphasis is placed upon the development of transformsthat provide consistently accurate reconstruction of quantizedsatellite and aerial reconnaissance images. The development oftransforms that preserve the intelligence that can be gained fromhighly compressed images transmitted over a limited bandwidth is ofdefense and security interest. This dissertation assembles a database ofpublicly available satellite images collected for a wide range ofsubjects, including aircraft and airfields, naval bases, army bases,cities, and factories. Experiments employing these images aregeared toward the development of filters appropriate for civilianand military aerial reconnaissance applications requiring limitedbandwidth for image transmission. Because the evolution employs theDWT algorithm, the resulting filters are easy to implement inhardware appropriate for digital signal processing applications.

This thesis shows how to incorporate Unified Modeling Language sequence diagrams into genetic algorithm-based automated test case generators to increase the code coverage of their resulting test cases. Automated generation of test data through evolutionary testing was proven feasible in prior research studies. In those previous investigations, the metrics used for determining the test generation method effectiveness were the percentages of testing statement and branch code coverage achieved. However, the code coverage realized in those preceding studies often converged at suboptimal percentages due to a lack of guidance in conditional statements. This study compares the coverage percentages of 16 different Java programs when test cases are automatically generated with and without incorporating associated UML sequence diagrams. It introduces a tool known as the Evolutionary Test Case Generator, or ETCG, an automatic test case generator based on genetic algorithms that provides the ability to incorporate sequence diagrams to direct the heuristic search process and facilitate evolutionary testing. When the generator uses sequence diagrams, the resulting test cases showed an average improvement of 21% in branch coverage and 8% in statement coverage over test cases produced without using sequence diagrams.

Present day air traffic control is reaching its operational limits and accommodating future traffic growth will be a challenging task for air traffic service providers and airline operators. Free Flight is a proposed transition from a highly-structured and centrally-controlled air traffic system to a self-optimised and highly-distributed system. In Free Flight, pilots will have the flexibility of real-time trajectory planning and dynamic route optimisation given airspace constraints (traffic, weather etc.). A variety of advanced air traffic management (ATM) concepts are proposed as enabling technologies for the realisation of Free Flight. Since these concepts can be exposed to unforeseen and challenging scenarios in Free Flight, they need to be validated and evaluated in order to implement the most effective systems in the field. Evaluation of advanced ATM concepts is a challenging task due to the limitations in the existing scenario generation methodologies and limited availability of a common platform (air traffic simulator) where diverse ATM concepts can be modelled and evaluated. Their rigorous evaluation on safety metrics, in a variety of complex scenarios, can provide an insight into their performance, which can help improve upon them while developing new ones. In this thesis, I propose a non-propriety, non-commercial air traffic simulation system, with a novel representation of airspace, which can prototype advanced ATM concepts such as conflict detection and resolution, airborne weather avoidance and cockpit display of traffic information. I then propose a novel evolutionary computation methodology to algorithmically generate a massive number of conflict scenarios of increasing complexity in order to evaluate conflict detection algorithms. I illustrate the methodology in detail by quantitative evaluation of three conflict detection algorithms, from the literature, on safety metrics. I then propose the use of data mining techniques for the discovery of interesting relationships, that may exist implicitly, in the algorithm's performance data. The data mining techniques formulate the conflict characteristics, which may lead to algorithm failure, using if-then rules. Using the rule sets for each algorithm, I propose an ensemble of conflict detection algorithms which uses a switch mechanism to direct the subsequent conflict probes to an algorithm which is less vulnerable to failure in a given conflict scenario. The objective is to form a predictive model for algorithm's vulnerability which can then be included in an ensemble that can minimise the overall vulnerability of the system. In summary, the contributions of this thesis are: 1. A non-propriety, non-commercial air traffic simulation system with a novel representation of airspace for efficient modelling of advanced ATM concepts. 2. An Ant-based dynamic weather avoidance algorithm for traffic-constrained enroute airspace. 3. A novel representation of 4D air traffic scenario that allows the use of an evolutionary computation methodology to evolve complex conflict scenarios for the evaluation of conflict detection algorithms. 4. An evaluation framework where scenario generation, scenario evaluation and scenario evolution processes can be carried out in an integrated manner for rigorous evaluation of advanced ATM concepts. 5. A methodology for forming an intelligent ensemble of conflict detection algorithms by data mining the scenario space.

Objectives: The main objective of the current dissertation is to establish a formal shape optimization procedure for a given bypass grafts end-to-side distal anastomosis (ETSDA). The motivation behind this dissertation is that most of the previous ETSDA shape optimization research activities cited in the literature relied on direct optimization approaches that do not guaranty accurate optimization results. Three different ETSDA models are considered herein: The conventional, the Miller cuff, and the hood models. Materials and Methods: The ETSDA shape optimization is driven by three computational objects: a localized collocation meshless method (LCMM) solver, an automated geometry pre-processor, and a genetic-algorithm-based optimizer. The usage of the LCMM solver is very convenient to set an autonomous optimization mechanism for the ETSDA models. The task of the automated pre-processor is to randomly distribute solution points in the ETSDA geometries. The task of the optimized is the adjust the ETSDA geometries based on mitigation of the abnormal hemodynamics parameters. Results: The results reported in this dissertation entail the stabilization and validation of the LCMM solver in addition to the shape optimization of the considered ETSDA models. The LCMM stabilization results consists validating a custom-designed upwinding scheme on different one-dimensional and two-dimensional test cases. The LCMM validation is done for incompressible steady and unsteady flow applications in the ETSDA models. The ETSDA shape optimization include single-objective optimization results in steady flow situations and bi-objective optimization results in pulsatile flow situations. Conclusions: The LCMM solver provides verifiably accurate resolution of hemodynamics and is demonstrated to be third order accurate in a comparison to a benchmark analytical solution of the Navier-Stokes. The genetic-algorithm-based shape optimization approach proved to be very effective for the conventional and Miller cuff ETSDA models. The shape optimization results for those two models definitely suggest that the graft caliber should be maximized whereas the anastomotic angle and the cuff height (in the Miller cuff model) should be chosen following a compromise between the wall shear stress spatial and temporal gradients. The shape optimization of the hood ETSDA model did not prove to be advantageous, however it could be meaningful with the inclusion of the suture line cut length as an optimization parameter.

We are continuously challenged by ever increasing problem complexity and the need to develop algorithms that can solve complex problems and solve them within a reasonable amount of time. Modularity is thought to reduce problem complexity by decomposing large problems into smaller and less complex subproblems. In practice, introducing modularity into evolutionary algorithm representations appears to improve search performance; however, how and why modularity improves performance is not well understood. In this thesis, we seek to better understand the effects of modularity on search. In particular, what are the effects of module creation on the search space structure and how do these structural changes affect performance? We define a theoretical and empirical framework to study modularity in evolutionary algorithms. Using this framework, we provide evidence of the following. First, not all types of modularity have an effect on search. We can have highly modular spaces that in essence are equivalent to simpler non-modular spaces. This is the case, because these spaces achieve higher degree of modularity without changing the fundamental structure of the search space. Second, for the cases when modularity actually has an effect on the fundamental structure of the search space, if left without guidance, it would only crowd and complicate the space structure resulting in a harder space for most search algorithms. Finally, we have the case when modularity not only has an effect in the search space structure, but most importantly, module creation can be guided by problem domain knowledge. When this knowledge can be used to estimate the value of a module in terms of its contribution toward building the solution, then modularity is extremely effective. It is in this last case that creating high value modules or low value modules has a direct and decisive impact on performance. The results presented in this thesis help to better understand, in a principled way, the effects of modularity on search. Better understanding the effects of modularity on search is a step forward in the larger issue of evolutionary search applied to increasingly complex problems.

This dissertation deals with the evolutionary optimization of ART neural network architectures. ART (adaptive resonance theory) was introduced by a Grossberg in 1976. In the last 20 years (1987-2007) a number of ART neural network architectures were introduced into the literature (Fuzzy ARTMAP (1992), Gaussian ARTMAP (1996 and 1997) and Ellipsoidal ARTMAP (2001)). In this dissertation, we focus on the evolutionary optimization of ART neural network architectures with the intent of optimizing the size and the generalization performance of the ART neural network. A number of researchers have focused on the evolutionary optimization of neural networks, but no research has been performed on the evolutionary optimization of ART neural networks, prior to 2006, when Daraiseh has used evolutionary techniques for the optimization of ART structures. This dissertation extends in many ways and expands in different directions the evolution of ART architectures, such as: (a) uses a multi-objective optimization of ART structures, thus providing to the user multiple solutions (ART networks) with varying degrees of merit, instead of a single solution (b) uses GA parameters that are adaptively determined throughout the ART evolution, (c) identifies a proper size of the validation set used to calculate the fitness function needed for ART's evolution, thus speeding up the evolutionary process, (d) produces experimental results that demonstrate the evolved ART's effectiveness (good accuracy and small size) and efficiency (speed) compared with other competitive ART structures, as well as other classifiers (CART (Classification and Regression Trees) and SVM (Support Vector Machines)). The overall methodology to evolve ART using a multi-objective approach, the chromosome representation of an ART neural network, the genetic operators used in ART's evolution, and the automatic adaptation of some of the GA parameters in ART's evolution could also be applied in the evolution of other exemplar based neural network classifiers such as the probabilistic neural network and the radial basis function neural network.

The generalization abilities of machine learning algorithms often depend on the algorithms’ initialization, parameter settings, training sets, or feature selections. For instance, SVM classifier performance largely relies on whether the selected kernel functions are suitable for real application data. To enhance the performance of individual classifiers, this dissertation proposes classifier fusion models using computational intelligence knowledge to combine different classifiers. The first fusion model called T1FFSVM combines multiple SVM classifiers through constructing a fuzzy logic system. T1FFSVM can be improved by tuning the fuzzy membership functions of linguistic variables using genetic algorithms. The improved model is called GFFSVM. To better handle uncertainties existing in fuzzy MFs and in classification data, T1FFSVM can also be improved by applying type-2 fuzzy logic to construct a type-2 fuzzy classifier fusion model (T2FFSVM). T1FFSVM, GFFSVM, and T2FFSVM use accuracy as a classifier performance measure. AUC (the area under an ROC curve) is proved to be a better classifier performance metric. As a comparison study, AUC-based classifier fusion models are also proposed in the dissertation. The experiments on biomedical datasets demonstrate promising performance of the proposed classifier fusion models comparing with the individual composing classifiers. The proposed classifier fusion models also demonstrate better performance than many existing classifier fusion methods. The dissertation also studies one interesting phenomena in biology domain using machine learning and classifier fusion methods. That is, how protein structures and sequences are related each other. The experiments show that protein segments with similar structures also share similar sequences, which add new insights into the existing knowledge on the relation between protein sequences and structures: similar sequences share high structure similarity, but similar structures may not share high sequence similarity.

The static elastic moduli of pavement layers can be considered to be among the most controversial physical properties in pavement engineering. In addition, pavement analysis using the static elastic moduli of the constituting layers is widely known and accepted by engineers and practitioners due to its simplicity. Nondestructive tests are commonly performed on existing pavements to measure the surface deflections, which in turn are used to backcalculate the elastic moduli of the pavement layers. However, the accuracy of the backcalculated moduli is dependent on the backcalculation procedure and the associated seed moduli. None of the existing classical backcalculation methods can find the “actual” pavement moduli due to the theoretical limitations of the existing methods. These limitations include the convergence to local optima due to the use of seed moduli, which in turn lead to erroneous pavement moduli. The genetic algorithms can be used to optimize the search domain of the backcalculated moduli to avoid the premature convergence to local optima. The use of genetic algorithms in pavement engineering is new and no guidelines or thorough investigations have been carried out to address all the aspects and challenges associated with the backcalculation procedure using the genetic algorithms. This study can be considered as the first comprehensive work that deals with all aspects of both pavement and genetic algorithms and how to merge them. In addition, this work can be considered as the first state of the art work on the backcalculation of pavement moduli using genetic algorithms. In this study, the use of genetic algorithms has been studied thoroughly to address all the important parameters and operators that affect the backcalculation process. In addition, recommendations and findings regarding all the details needed to carry out the backcalculation process were identified and discussed thoroughly. New novel methods to study the interaction between the genetic operators and parameters and their effect on the backcalculation process were developed. Recommendations regarding the genetic operators as well as the genetic parameters were presented throughout the work. In addition, the AASHTO recommended ranges of pavement moduli were modified based on the study results to suit the GAs backcalculation process. On the other hand, a new novel method was developed to automate the backcalculation process. The automation of the backcalculation process was aimed at reducing the number of inputs needed to carry out the backcalculation process and to make it more appealing to be used in practice. A new Dynamic Parameterless Genetic Algorithm (DPGA) was developed as part of this work. The new DPGA can be extended to many other applications of genetic algorithms including robotics and optimizations. A new program (BackGenetic3d) was developed based on the novel MultiSmart3D program developed by the Computation and Simulation Group at the University of Akron. The new program is the first in the world that can backcalculate the pavement moduli of pavement systems with any arbitrary number of layers, loading conditions, and loading configurations. Existing classical programs use backcalculation procedures that lead to local optima and limited to a maximum of 5 pavement layers and one loading circle with uniform pressure.

This dissertation provides the theory, design, and implementation of a cognitive engine, the enabling technology of cognitive radio. A cognitive radio is a wireless communications device capable of sensing the environment and making decisions on how to use the available radio resources to enable communications with a certain quality of service. The cognitive engine, the intelligent system behind the cognitive radio, combines sensing, learning, and optimization algorithms to control and adapt the radio system from the physical layer and up the communication stack. The cognitive engine presented here provides a general framework to build and test cognitive engine algorithms and components such as sensing technology, optimization routines, and learning algorithms. The cognitive engine platform allows easy development of new components and algorithms to enhance the cognitive radio capabilities. It is shown in this dissertation that the platform can easily be used on a simulation system and then moved to a real radio system. The dissertation includes discussions of both theory and implementation of the cognitive engine. The need for and implementation of all of the cognitive components is strongly featured as well as the specific issues related to the development of algorithms for cognitive radio behavior. The discussion of the theory focuses largely on developing the optimization space to intelligently and successfully design waveforms for particular quality of service needs under given environmental conditions. The analysis develops the problem into a multi-objective optimization process to optimize and trade-of of services between objectives that measure performance, such as bit error rate, data rate, and power consumption. The discussion of the multi-objective optimization provides the foundation for the analysis of radio systems in this respect, and through this, methods and considerations for future developments. The theoretical work also investigates the use of learning to enhance the cognitive engine's capabilities through feed-back, learning, and knowledge representation. The results of this work include the analysis of cognitive radio design and implementation and the functional cognitive engine that is shown to work in both simulation and on-line experiments. Throughout, examples and explanations of building and interfacing cognitive components to the cognitive engine enable the use and extension of the cognitive engine for future work.

Antenna design has grown more stringent and difficult over the years as the world becomes strictly a wireless environment. The inherent tradeoffs that exist between gain, radiation pattern, bandwidth, and physical size and the multiple parameters that must be considered make antenna design a lengthy and tedious process. Methods have been devised which automate this complex process of antenna optimization through the use of genetic algorithms, particle swarm optimization, and simulated annealing. Genetic algorithms are capable of handling a large number of design parameters and work for optimization problems that have discontinuous or non-differentiable multi-dimensional solution spaces, making them ideal for antenna optimization. In the present work, a genetic algorithm has been used for size reduction in microstrip patch antennas and design tradeoff optimization between beamwidth and gain in helical antennas. A method for reducing the size of microstrip patch antennas by up to 75% by removing rectangular and circular slots from the metal of the microstrip patch is presented. A solid patch antenna that resonates at 10 GHz is forced to resonant at 6 GHz through the removal of the different shaped slots. Given the number and shape of the slots, the genetic algorithm is used to optimize the size and location of the slot on the patch. The designs are obtained by interfacing the genetic algorithm and Ansoft High Frequency System Simulator (HFSS) and validated through design, construction, and testing. High gain, with broad half-power beamwidths (HPBW) is traditionally extremely difficult to achieve due to the inherent tradeoff between the two. A genetic algorithm has been applied to design a helical antenna with a gain of 10 dB and HPBWs of 60 degrees. In order to achieve this, three physical parameters of the helix have been changed, namely the pitch, helix radius, and the ground plane geometry. The second objective is to create an antenna that displays different HPBWs in the two radiation planes. This could be extremely useful in many communication environments and there is yet no existing method to achieve this. The genetic algorithm produced a helical antenna that shows a 19 degrees difference in HPBW between the two radiation planes, while still displaying a 7 dB gain and low side lobes. Numerical Electromagnetic Code 4 (NEC4) is used, and a method of communication between MATLAB and NEC4 has been developed to make the genetic algorithm optimization possible.

This research work deals with the design and optimization of a large composite structure. In design of large structural systems, it is customary to divide the problem into many smaller independent/semi-independent local design problems. For example, the wing structure design problem is decomposed into several local panel design problem. The use of composite necessitates the inclusion of ply angles as design variables. These design variables are discrete in nature because of manufacturing constraint. The multilevel approach results into a nonblended solution with no continuity of laminate layups across the panels. The nonblended solution is not desirable because of two reasons. First, the structural integrity of the whole system is questionable. Second, even if there is continuity to some extent, the manufacturing process ends up being costlier. In this work, we develop a global local design methodology to design blended composite laminates across the whole structural system. The blending constraint is imposed via a guide based approach within the genetic algorithm optimization scheme. Two different blending schemes are investigated, outer and inner blending. The global local approach is implemented for a complex composite wing structure design problem, which is known to have a strong global local coupling. To reduce the computational cost, the originally proposed local one dimensional search is replaced by an intuitive local improvement operator. The local panels design problem arises in global/local wing structure design has a straight edge boundary condition. A postbuckling analysis module is developed for such panels with applied edge displacements. A parametric study of the effects of flexural and inplane stiffnesses on the design of composite laminates for optimal postbuckling performance is done. The design optimization of composite laminates for postbuckling strength is properly formulated with stacking sequence as design variables. Next, we formulate the stacking sequence design (fiber orientation angle of the layers) of laminated composite flat panels for maximum supersonic flutter speed and maximum thermal buckling capacity. The design is constrained so that the behavior of the panel in the vicinity of the flutter boundary should be limited to stable limit cycle oscillation. A parametric study is carried out to investigate the tradeoff between designs for thermal buckling and flutter. In an effort to include the postbuckling constraint into the multilevel design optimization of large composite structure, an alternative cheap methodology for predicting load paths in postbuckled structure is presented. This approach being computationally less expensive compared to full scale nonlinear analysis can be used in conjunction with an optimizer for preliminary design of large composite structure with postbuckling constraint. This approach assumes that the postbuckled stiffness of the structure, though reduced considerably, remains linear. The analytical expressions for postbuckled stiffness are given in a form that can be used with any commercially available linear finite element solver. Using the developed approximate load path prediction scheme, a global local design approach is developed to design large composite structure with blending and local postbuckling constraints. The methodology is demonstrated via a composite wing box design with blended laminates.

The application of Visible and Near-Infrared (VIS/NIR) spectroscopy to determine economic adulteration of crabmeat was determined. Crabmeat samples were adulterated in 10% increments according to weight. The adulterants chosen were surimi-based imitation crabmeat, due to its low cost and availability, and blue swimmer crabmeat, the most prevalent type of crabmeat imported into the United States. Several data pre-treatments and different chemometric analyses, Partial Least Squares (PLS), Principal Component Regression (PCR), and Multiple Linear Regression (MLR), were investigated to determine the predictive ability of VIS/NIR spectroscopy in detecting economic adulteration and species authenticity. In addition, wavelength variables selected by a genetic algorithm were evaluated to improve predictive ability of economic adulteration of crabmeat. Absorption spectra of adulterated samples were dominated by water overtones. Absorption decreased for increasing level of adulteration. PLS was favored over PCR due to its predictive ability and lower number of latent variables used in model development. The first derivative data pre-treatment generated the best results, though similar results were gathered with the untreated data. First derivative data, using data from a correlogram, generated the best PLS model to detect economic adulteration of crabmeat adulterated with surimi-based imitation crabmeat. Economic adulteration percentage was predicted within 2.5%. The second derivative data generated the highest errors and, thus, was not deemed an appropriate pre-treatment method. For samples adulterated with blue swimmer, the first derivative data also generated the best results utilizing the full spectrum with a Standard Error of Calibration (SEC) and Standard Error of Prediction (SEP) of 5.64. Using a partitioned spectrum, however, the second derivative data performed better (SEC and SEP of 4.91 and 5.17, respectively). Regardless of the data pre-treatment, VIS/NIR spectroscopy was able to detect species authenticity and economic adulteration to less than 6% for samples adulterated with blue swimmer crabmeat. Variable selection via genetic algorithm enabled MLR to detect economic adulteration, making this the best overall method due to its low SEC and SEP (4.21 and 3.98, respectively). All of these findings provide a baseline for the design and development of a fast, reliable, and accurate technology for on-line detection of economic adulteration of crabmeat.