Search

Applications of bio-inspired algorithms to feature selection offering promising performance can be frequently encountered in the literature. Serving as wrappers, their application essentially boils down to the selection of a classifier and an appropriate metric for evaluating model performance. Although the choice of classifier was investigated to a viable extent, this is, however, not the case for the choice of metric and its impact on performance. Nevertheless, the utilisation of a wide variety of metrics is clearly apparent, albeit to different proportions. This raises the question of whether some metrics might be a better choice than others for bio inspired wrapper-based methods. This paper sheds some light on this matter by comparing different metrics from different perspectives, like correlation, general performance and subset sizes. Unexpected results were obtained, to say the least, apart from some minor exceptions. Generally, no single metric appears to have an edge over the others. Yet, further investigation is certainly warranted.

This paper proposes a novel differential evolution operator to enhance the rational exploration of the multi-objective optimization based on decomposition for solving the extended environmental/economic dispatch problem. In practical power engineering, besides the fuel cost and harmful emission become the major concerns, transmission loss does as well simultaneously. Meanwhile, the increasing near-real-time decisions push the conventional differential evolution as a solver to promote the solving speed. Therefore, a dynamic response surface fitting between the control vector and the fitness value is introduced for getting the approximation relationship that can provide a sequence of rational searching directions to accelerate the solving process to approach the optima of the 3-objective optimization. The proposed approach shows promising performance while simulated by using DTLZ1-4 and IEEE-30 standard power systems.

Low switching frequency modulation techniques such as Selective Harmonic Elimination - Pulse Width Modulation (SHE-PWM) or Optimal Pulse Pattern (OPP) are commonly used in medium voltage - high power multilevel converters to improve their efficiency. These techniques require solving non-linear equations in order to calculate the firing angles. These equations can be solved by meta-heuristic optimization algorithms, which depend on a set of hyperparameters that must be properly adjusted to ensure their effectiveness. This paper presents a methodology to tune different meta-heuristic algorithms to fairly compare them and select the best algorithm to solve SHE-PWM/OPP formulation. The methodology is applied considering different meta-heuristic algorithms such as genetic algorithm, differential evolution, harmony search and simulated annealing. It has been validated in SHE-PWM and OPP techniques with different number of firing angles.

In order to solve the problem of low efficiency when the interleaved parallel DC-DC converters adopting current-sharing scheme work in low load conditions, this paper proposes an efficiency optimization control method which finds the load current distribution scheme for optimal efficiency by applying differential evolution (DE) algorithm and selects appropriate operation mode according to the system state. Furthermore, to test dynamic performance of different control strategies, simulations and experiments are carried out with both Dual-PI control and modified PI-PMC control based on Two-phase interleaved synchronous rectification BUCK converter. The results indicate efficiency optimization control method proposed in this paper can greatly improve the efficiency and accelerate dynamic response when adopting PI-MPC control.

One of the reason that hinders ridesharing from acceptance by general public is due to the safety concern to share a vehicle. Recently, the problem to meeting trust requirements in ridesharing systems has been formulated. As this problem is a complex discrete decision problem, it can be solved by metaheuristic algorithms. In the literature, one solution algorithm based on Particle Swarm Optimization (PSO) and another solution algorithm based on Firefly algorithm were developed to solve this problem. The goal of this study is to develop another solution algorithm based on Differential Evolution (DE) approach and compare the proposed algorithm with PSO and Firefly algorithms. Several mutation strategies in DE approach have been implement to solve the optimization problems. We evaluate the DE based algorithms by conducting experiments. We analyze the results to illustrate effectiveness of the DE based algorithms.

Various optimization objective functions have been proposed for harmonic optimization problems to eliminate unwanted specific harmonic or reduce total harmonic distortion (THD). However, the objective functions are usually intuitively selected, which lack a theoretical basis. In this paper, five objective functions with different mathematical forms are evaluated by two commonly used bio-inspired optimization algorithms, namely the particle swarm optimizers (PSO), and differential evolution (DE). By comparing the THDs, the low-order individual harmonics, and the computational time, resulted from the five objective functions, some guidelines are summarized for selecting objective functions for the bio-inspired harmonic optimization algorithms.

Intelligent optimization algorithms, such as the genetic algorithm (GA) and particle swarm optimization (PSO), have been widely used for harmonic minimization in power converters. However, these algorithms usually evaluate the fitness function for hundreds of or even more populations, which leads to a huge computing burden and memory consumption. Hence, in practical applications, it is difficult to real-time implement these algorithms on the traditional central processing units (CPUs). On the other hand, although the population size is huge, each of them performs the same calculations independently, which is very suitable for parallelization on the graphical processing units (GPUs). In this paper, a parallel version of differential evolution (DE) on GPUs is proposed. Compared to the traditional CPU-based DE algorithm, the GPU-based parallel DE algorithm executes hundreds of times faster in solving the harmonic minimization problem. Some computational results of 21 switching angles for three-level inverters are given. Also, some guidelines for the parameter selection, such as the population size, the grid allocation, etc., on algorithm parallelization are discussed and summarized.

Parameter extraction is a challenging task, as it searches for a solution inside a high dimensional plus non-convex space. To be able to apply well known gradient based optimizers, the problem is dissected into multiple simpler yet intertwined tasks, which yields a complex and manual labour intensive procedure. On the contrary to gradient based methods, genetic algorithms perform excellent on global search problems, which eliminates the need for such a sophisticated workflow. In this paper a highly automated methodology is presented that is capable of replacing the standard manual extraction sequence for the BSIM MOSFET compact model. Due to its superior extreme finding behaviour, the Differential Evolution algorithm is applied in combination with a special error metric to ensure a high fitting quality, in all regions of the output and transfer curves. Repeatably good results for 20k measurement points are obtained, with a reduction of factor 10 in total fitting duration, while coincidentally consuming mostly computation instead of manual labour time.

Marie Skłodowska-Curie Action Project name: Optimized Dynamic Point Cloud Compression, acronym: OPT-PCC, grant number 836192 The file attached to this record is the author's final peer reviewed version. The Publisher's final version can be found by following the DOI link. In video-based point cloud compression (V-PCC), one geometry video and one color video are generated from a dynamic point cloud. Then, the two videos are compressed independently using a state-of-the-art video coder. In the Moving Picture Experts Group (MPEG) V-PCC test model, the quantization parameters for a given group of frames are constrained according to a fixed offset rule. For example, for the low-delay configuration, the difference between the quantization parameters of the first frame and the quantization parameters of the following frames in the same group is zero by default. We show that the rate-distortion performance of the V-PCC test model can be improved by lifting this constraint and considering the rate-distortion optimization problem as a multi-variable constrained combinatorial optimization problem where the variables are the quantization parameters of all frames. To solve the optimization problem, we use a variant of the differential evolution algorithm. Experimental results for the low-delay configuration show that our method can achieve a Bj{\o}ntegaard delta bitrate of up to -43.04% and more accurate rate control (average bitrate error to the target bitrate of 0.45% vs. 10.75%) compared to the state-of-the-art method, which optimizes the rate-distortion performance subject to the test model default offset rule. We also show that our optimization strategy can be used to improve the rate-distortion performance of two-dimensional video coders.

A power system network has dynamic and complex characteristics. The continuity of electricity supply to customer side from the generation in stable, secure, and reliable is being a top priority to be considered. Preserving an operation of a power system on its operating point depends on its protection system scheme. Furthermore, carrying out the protection system requires a relay setting that can coordinate to carry out security. However, calculating the time dial setting (TDS) and the pickup current on each relay for practical work in industry is more difficult, complex, and takes a long time. In this paper, the directional overcurrent relay (DOCR) protection is coordinated by using hybrid particle swarm optimization - differential evolution (PSO-DE) algorithm. The proposed method is to minimize the TDS time and operating time of the relays by satisfying the protection coordination requirements. The simulation results showed that the total operating time and TMS using PSO-DE have faster operating times and TDS than other optimization algorithms.

The development of renewable energy has become massive, especially the implementation of solar panels. Even so, solar panels highly depends on irradiation levels and temperature conditions. The partial shading condition is also affects the solar panel output power. In order to get the solar panel maximum power, MPPT method can be used. This research suggests MPPT method with Self Adaptive Differential Evolution (SADE) algorithm. The PV system uses 3 large 100 Wp solar panels connected in series and a buck converter. The MPPT SADE system uses PSIM software so to test the MPPT SADE performance in order to get the maximum power output from PV Array and compared to the HPO and PSO algorithm. The results are shown that by MPPT SADE is able to do GMPP tracking in partial shading condition with average tracking accuracy and tracking period 99.99% and 0.24 seconds.

Leakages in Water Distribution Networks cause significant economical losses as well as environmental damage. Therefore, this paper proposes a novel leak location approach. The proposed method consists in using a modified differential evolution (DE) algorithm to solve the leak location task as an inverse problem based on the network’s hydraulic model. The modified DE algorithm considers the topological characteristics of the network. Temporal reasoning of the results is integrated with the proposal to improve the leak location performance. The strategy was tested under different uncertainty conditions with the model of the Modena water distribution network showing a performance ranging from 94 to 98% leak location accuracy with reduced zone sizes compared to previous approaches based on data-driven classification tools.

Under the environment of the new round of power system reform, the accounting mode of power grid is changed from price difference to cost plus income, and the competition on the power selling side is released, which greatly affects the income of regional power grids, thus greatly reducing the profit margin of power grid enterprises and greatly restricting the investment capacity. The new power reform also proposes to strengthen the overall planning of power. The investment decision-making problem of new power system is a high-dimensional, nonconvex and multi-constrained optimization problem, and the integration of wind farms further increases the difficulty of the problem. In order to optimize the problem better and enhance the convergence performance of the algorithm, based on DE(Differential Evolution) algorithm, some improvement measures such as shared fitness and adaptive adjustment of control parameters are introduced. The research results show that the improved algorithm proposed in this paper improves the convergence speed of DE algorithm, shortens the operation time to a certain extent, and obtains better optimization results, which verifies the effectiveness of the method.

In this work, the optimal design of an unmanned aerial vehicle (UAV) wing spar by multi-objective optimization genetic algorithms is studied. An electromechanical finite element model (FEM) for piezolaminated bimorph cantilever structure with embedded piezoceramics is used combined with multiobjective genetic algorithms. The FEM formulation is based on laminated plate theory combined with the first-order shear deformation theory (FSDT) for which each piezoelectric layer has one additional electrical degree of freedom. Non-dominated Sorting Genetic Algorithm II (NSGA-II), Non-dominated Sorting Genetic Algorithm III (NSGA-III) and Generalized Differential Evolution 3 (GDE3) algorithm have been used to optimize the geometric and electric circuit parameters of a UAV generator for maximum power output and minimum mass added by the embedded piezoceramics. It is shown that the proposed algorithms are effective in developing optimal Pareto front curves for maximum electrical power output of the generator spar and minimum mass add by the embedded piezoceramics. Comparison of the generated solutions on the Pareto Front for UAV wing spar design shows that all three algorithms achieve solutions that closely resemble a constructed reference Pareto optimal front.

Strain and temperature are the key measurement parameters in a distributed optical fiber sensing system based on Brillouin scattering, which may be determined by analyzing the Brillouin center frequency shift. In order to accurately extract the features of the Brillouin scattering spectrum of a distributed sensing system, we propose a high accurate fitting algorithm using differential evolution (DE) algorithm to optimize the least squares support vector machine (LSSVM) algorithm, and compare it with traditional Levenberg-Marquardt (LM) and the five times polynomial curve algorithm. The results show that the DE-LSSVM algorithm has higher fitting degree and smaller frequency shift error for different line widths, signal-to-noise ratio and linear weight ratio. Therefore, this algorithm can effectively improve the accuracy of frequency shift parameters in distributed optical fiber sensing system based on Brillouin scattering.

Optimal test sequence problem of fault diagnosis is an NP-complete problem, which inspires lots of studies. However, the multi-fault diagnosis of that is more critical and more challenging due to the overlapped signatures and the masking false failures, but this problem is very lacking in research. We propose a novel diagnostic strategy based on the major-minor testing sequences instead of the traditional fixed testing sequences. The major testing sequences lead the whole diagnosis which are evolved by an improved differential evolution algorithm (Ensemble of differential evolution variants, EDEV). Aiming at the different source faults, different minor testing sequences are generated from the major testing sequence by heuristic search algorithms. With the cooperation of these two sequences, the expected testing costs are reduced without losing the fault isolation rate. As for the hidden and masking false failures, we defined the concepts of the parent fault and fault link, so that the complexity of calculating the masking false failure is reduced from O(2mn) to O(n2), and the process of isolating multiple faults is more efficient. The study cases from real world show that the method is valid, and it is able to solve both the single-fault and multi-fault diagnosis effectively.

Evolutionary algorithms are effective tools for solving different problems such as mathematical modelling. The paper proposes an evolutionary approach to the identification of dynamic systems. It combines a genetic programming algorithm and a differential evolution method. The main modifications of the evolutionary steps for these algorithms are presented. The investigation and testing of the proposed approach was performed on problems of various dimensions.

Decision tree learning algorithms are quite old but very popular because of their efficiency. Decision trees are built sequentially from node to node, resulting in a "near-optimal" tree. This article discusses an approach to training a decision tree based on relearning already built nodes, i.e. a tree is not built sequentially, but with a return to the previous nodes. Such algorithms as Separation Measure and Differential Evolution are applied for this approach. Separation Measure selects the attributes to optimize the thresholds, and Differential Evolution performs optimization. The considered approach to learning a decision tree is compared with standard algorithms. Statistical analysis is performed to show the reliability of the results, with Student’s t-test being applied in this case. On average, the efficiency of the standard and modified decision tree learning algorithms are at the same level. However, a modified decision tree learning algorithm can find a tree that is more efficient than the trees built by the standard algorithm.

The possibility of enhancing the performance of the Probit classification model by adding a game theoretic flavour to the model parameters is explored in this paper. A game in which each attribute of the data set represents a player choosing its parameter is devised. The Nash equilibrium of this game is approximated by using a Differential Evolution algorithm. Numerical examples illustrate the potential of this approach.

Prolonging sensor and network life is a significant concern for wireless sensor networks and in this paper we present a novel optimization technique for determining the superframe slot allocation for a beacon-enabled 802.15.4 superframe operating with sensors with different data acquisition periods. The optimization approach is based on a differential evolution algorithm that leverages a flat cluster network topology to constrain the contention free period (CFP) and the contention access period (CAP) of the superframe. In this work we also define a multi-superframe structure to help conserve energy in the network. A sensitivity analysis of the optimization shows that the optimized results are not too sensitive to the clustering algorithm but are very sensitive to the original network configuration, such as number of nodes and their locations and transmission range, and therefore is better suited for static sensor networks. We also demonstrate that the optimized superframe results in a 5% gain in energy conservation over other approaches that optimize the superframe based on the ratio of the beacon order (BO) to superframe order (SO).

Massive MIMO provides a high spatial degree of freedom (DoF) to support multiple users. To gain the spatial DoF, accurate channel information is crucial, and the pilot is important to obtain it. However, in grant-free transmissions, different users may select the same pilot, which leads to pilot collisions. To solve this problem, this paper proposes an irregular superimposed multi-pilot. The proposed superimposed structure provides a large multi-pilot pool than the conventional cascaded one. Using this structure, the number of single-pilots in multi-pilot can be different, and the pilot number distribution is optimized via a novel Monte-Carlo-based differential evolution. The simulation results verify the analysis and show that the proposed method outperforms existing works.

Constant development of electric power systems leads to their constant enlargement and complication; new ways of their control appear. In this regard, the existing models and software for adequacy assessment may work defective and ineffectively from the point of view of the adequacy of the obtained results, as well as the speed and accuracy of calculations. The key role in adequacy assessment of electric power systems (EPS) are played by optimization methods that allow to correctly determine the minimum of power shortage that occurs in various states of the EPS. A review of modern computational systems for adequacy assessment showed that the general concept of mathematical models is the same and can be described within the framework of the flow distribution problem. Despite this, each mathematical model is unique in its own way and requires an individual approach to its optimization. The purpose of this work is to analyze the efficiency of calculations in terms of accuracy and speed of various versions of the differential evolution (DE) method for the specified mathematical models within the framework of adequacy assessment of EPS. To achieve this goal, we solved several problems: two mathematical models were identified - a nonlinear model for minimizing the power shortage with the quadratic losses in flows and its modification with the controlled sections; differential evolution methods, including standard DE, composite DE, JDE, chaotic DE, adaptive DE; mutation strategies: DE/rand/1, DE/best/1, DE/rand/2, DE/best/2, DE/rand/3, DE/best/3, DE/current-to-rand/1, and DE/current-to-best/1. In this paper we tested the effectiveness of differential evolution methods, with different mutation strategies and different scales of EPSs. The experimental part was carried out using a software that was independent development by authors using C++. This complex includes the implementation of mathematical models and methods. The methods were tested on two systems with different numbers of adequacy zones, including those with three and seven adequacy zones. According to the research results, self-adaptive methods of DE are of the greatest interest for the further use and development of methods for this problem, due to automatic adjustment of the method parameters for each of the considered models and systems.

Forecasting country-wide energy demand is an important activity for determining and progressing towards an energy secure future. In this paper, one-year-ahead energy use for the country of Spain has been modeled using metaheuristics with socio-economic indicators as the input variables. A total of 14 input variables were selected (including gross domestic product, the population of the country, import and export quantities, etc.). The data for energy use as well as the 14 input variables have been collected from the public websites for 45 years (for the period of 1970–2014). The approach here is to bifurcate the data into training and testing datasets with equal number of data points. The training dataset is used to train the algorithm while the testing dataset is used to evaluate the performance of the algorithm on the independent data. The metaheuristics applied has two components: A highly recursive Grammatical Evolution (GE) and Differential Evolution (DE). GE identifies the model structure while the DE optimizes the coefficients of the models. The performance of the algorithm was evaluated in terms of various statistical errors (RMSE 0.0393, Avg. Error −0.0091, R2 0.9503, Absolute Error 0.0312, and Relative Error 0.0427). The results so obtained confirm that the proposed GE+DE method is highly accurate and can be successfully used to model the country's future energy demand precisely.

The load of a power system usually presents a certain range of nonlinear fluctuation with time. Even then, the load characteristics still follow certain rules which can be exploited to optimise and improve the accuracy of computer-based Short-Term Load Forecasting (STLF) models. Therefore, this paper presents a mGA (micro–Genetic Algorithm) embedded multi-population DE (Differential Evolution) to optimise an Artificial Neural Network (ANN) STLF model. Firstly, the mGA embedded multi-population DE is proposed, to improve and balance the global and local search. Then the proposed DE is applied to optimise the weights during the training of the ANN. The overall model’s performance is evaluated using publicly available Panama electricity load dataset against four state-of-the-art machine learning algorithms. The evaluation results show that the proposed DE based NN STLF model has higher prediction accuracy compared to the other selected machine learning algorithms.

Aiming at the problem of dynamic flexible job shop scheduling, a multi-step genetic algorithm is proposed. Firstly, special encoding and decoding methods are used according to the characteristics of the problem; then, introducing differential evolution mutation in the mutation process to improve the diversity of the population; at the same time, during the crossover process, in order to ensure the validity of the solution, the group was divided into two groups for pair wise crossover pairing using the random crossover method. Finally, aiming at the actual operation problems of enterprises, the dynamic scheduling in the event of a machine failure and the dynamic scheduling in the case of urgent orders are discussed separately. The test results show that the method proposed in this paper can get a good scheduling scheme and is an effective method for solving dynamic flexible job shop scheduling problems.

This paper proposes an improvement of genetic algorithm for optimization problems. In this study, the Rao algorithm was applied in crossover and mutation operators instead of traditional crossover and mutation. The algorithm was tested on six benchmark problems and compared with differential evolution (DE), JDE self-adaptive algorithm, and intersection mutation differential evolution (IMDE) algorithm. The computation results illustrated that the proposed algorithm can produce optimal solutions for three of six functions. Comparing to the other three algorithms, the proposed algorithm has provided the best results. The findings prove that the algorithm should be improved in this direction and show that the algorithm produces several solutions obtained by the previously published methods, especially for the continuous step function, the multimodal function and the discontinuous step function.

This paper proposes a new method of maximum power point tracking (MPPT) using novel asymmetric fuzzy MPPT controller (AFC-MPPT) for photovoltaic (PV) system improved from Perturbed and Observed (P&O) algorithm. This PV system is composed of the PV solar panel, buck-boost DC-DC converter and proposed AFC-MPPT controller. The proposed AFC-MPPT technique is fully compared to the other advanced MPPT algorithms such as classic fuzzy FC-MPPT, symmetric fuzzy FC-MPPT, P&O methods through which confirms the effectiveness and the superiority of the proposed asymmetric fuzzy AFC-MPPT technique. Furthermore, important input-output membership functions coefficients of AFC-MPPT controller will be optimally identified using modified differential evolution (mDE) algorithm. The simulation and experiment results are adequately presented to demonstrate the performance of AFC-MPPT method.

This paper proposes a new bipedal gait planning algorithm with natural Zero Moment Point (ZMP) references and preset foot-lifting magnitude. The building a model of walking pattern generation for biped robot based on model Nonlinear Autoregressive eXogenous combined with Multi-Layer Perceptron Neural Networks (NARX-MLPNN) is realized. In the NARX-MLPNN model, there are 4 inputs including 2 coordinate data of the reference ZMP point and 2 coordinate data of the actual ZMP point delayed in one step, there are 4 outputs which are gait generation parameters of biped robot. Furthermore, the Improved Differential Evolution (IDE) algorithm used to the identification the weights value of the MLPNN model, to biped walking with natural ZMP references and preset foot-lifting magnitude. It is convincingly shown that we can easily generate a biped walking gait with the efficiency and high performance. The validity of the generated patterns is confirmed by simulations adequate.

In this paper, we propose an efficient reference-point based surrogate-assisted multi-objective differential evolution algorithm for analog/RF circuit synthesis, which aims to greatly reduce simulation time while maintaining accuracy. First, offspring are generated by differential evolution operator and polynomial mutation. Then we predict the offspring by Gaussian process. Finally, sort the offspring by modified reference-point based NSGA-III criterion, only the most promising one instead of multiple designs is able to be simulated by the circuit simulator. Experimental results on two real-world circuits show that compared with two state-of-the-art multi-objective evolutionary algorithms, our method can achieve up to 6X simulation time reduction without surrendering any accuracy.

The Flexible Job Shop Scheduling Problem is well known as NP-hard and constrained. It is generic in the sense that many other problems are defined along similar lines. The Warehouse Resource Allocation problem belongs to this family, with its typically massive scale as an add-on factor. The severe non-concurrency constraint on both Tasks (Jobs) and Agents (Machines) remains valid. Here the problem is formulated in terms of the dual objectives of minimizing delay in completion of Tasks, and maximizing uniform utilization of available Agents. Differential Evolution (DE) is used for this multi-objective optimization, using both non-dominated sorting as well as variant linear combinations of multiple objectives. Efficacies of the two approaches are evaluated. The practical problem deals with thousands of Tasks represented as variables of the solution, which need to be loaded onto scores of Agents that appear as integer solutions of these variables. Since DE at basics is formulated to work on continuous spaces, here the integer solutions are mapped onto real spaces where DE is executed and then transformed back into integer space. Local Search is used to augment the baseline Global Search. A Divide-and-Conquer approach is implemented for the solution, with some novelty for handling variant, nonuniform sizes of different classes of Tasks and Agents.

Superdirective antenna can realize high channel capacity of multiple-input multiple-output system (MIMO). The optimization of superdirective scattering by a novel weighting optimizing goal and Dirac-Delta function model is proposed. Both main-lobe and side-lobe characteristics of the lobe are optimized. Compared with the traditional optimization method, which only takes the main-lobe characteristic into consideration, the directivity has improved by 50% (from 6.0 to 9.0). Meanwhile, the radius of the scattering object is only $0.1 \lambda_{0}$, which indicates this antenna is a super-directive antenna.

This paper proposes a new robust static output feedback control design considering the linear quadratic regulator (LQR) problem based on Linear Matrix Inequalities (LMIs). The output static feedback controller can be used when all the state variables are not available for feedback. This way, the controller design can solve practical problems in which there are no sensors for all state variables of the plant. Furthermore, the presented robust control methodology minimizes an upper bound of a quadratic index (guaranteed cost) related to the output and the control signal of the uncertain closed-loop linear system. Through the designer’s knowledge of the system to be controlled, it is possible to obtain optimized performances. In order to find the best guaranteed cost related to the performance of the system, an algorithm of differential evolution for global optimization was used. The controller was implemented in a bench scale earthquake simulator and the results illustrate the effectiveness of the proposed methodology. In the implementation, a signal fault is assumed, and even in the presence of fault occurrence, the oscillations are attenuated by the proposed robust control.

The Subspace Clustering Algorithm Based on Differential Evolution (DESC) performs well among the subspace clustering algorithms of global search. But the fixed number of iterations used may cause the algorithm to terminate when it does not converge or lead to overfitting. The improved algorithm proposed in this paper replaces the mutation operators of the DESC and uses the smoothness of the change in the optimal value of the objective function as the basis of judging convergence –– so that the algorithm can adaptively determine the termination timing and solve the above problems to a large extent. The experimental results show clearly that the proposed algorithm can terminate in the appropriate number of iterations in most cases and has certain advantages over the original algorithm in clustering accuracy and running time.

Urban growth modelling is a current trend in geo-computation due to its impact on the local living environment and the quality of life. The FUTure Urban-Regional Environment Simulation (FUTURES) model produces projections of landscape patterns by coupling land suitability, per-capita demand, and patch growing algorithm (PGA) sub-models. In particular, the PGA is the urban growing simulator component that takes into account the stochastic nature of urban development. It requires a set of parameters, namely compactness mean, compactness range, and discount factor to approximate the general characteristics of the urban development structure. The fitness of the parameters is measured by computing the difference between the area and compactness histograms of the observed and simulated urban growths. On the one hand, the authors find these parameters via an exhaustive grid search; nevertheless, this requires evaluating all the points in the grid, which implies a high computational cost because each point is associated with several PGA simulations. In addition, the approximation is limited to be one of the points in the grid. Thus, the better the precision is, the higher the required computational cost. On the other hand, evolutionary algorithms have been widely used for the automatic calibration of parameters but, in general, they are not designed to use a low number of evaluations, require expert tuning, and to define the stop criteria. Therefore, we propose an algorithm to find the adequate parameters, with minimum expert intervention, using an Estimation of Distribution Algorithm designed to use a low number of function evaluations (EDALNFE) when compared to the grid search. EDALNFE provides several assets: it delivers competitive results, it requires a low number of function evaluations, it does not require expert settings, and it is equipped with an automatic stop criterion. The proposed algorithm is compared to exhaustive grid search (the only method readily available in the FUTURES package) and differential evolution to demonstrate its superior performance.

This paper proposes an economic optimal scheduling method of based on the stackelberg game theory to address the energy trading of distribution network(DN) with multi-microgrid (MMG). First, the optimal scheduling process is designed as a one-leader N-follower stackelberg game. The equilibrium solution of the game is solved by the mixed integer programming (MIP) and differential evolution (DE) algorithms. The distribution network operator (DNO) is taken as the leader, and the microgrid operator (MGO) is regarded as a follower, aiming at the minimum operation cost. DNO encourages MGO to participate in energy trading by setting internal transaction price; MGO makes energy storage and load decision in response to internal transaction price, and then affects the price strategy of DNO through the net load. Moreover, the power flow of distribution network is constrained to ensure its safe and stable operation. Finally, an example analysis of the improved IEEE 33-bus system is designed to proof the effectiveness of the presented energy trading method.

Hunger games search (HGS) is a recently proposed meta-heuristic algorithm based on hunger-driven activities and animal behavior choices, and it has been proven to possess global exploration ability and can solve both constrained and unconstrained problems effectively. Differential evolution (DE) algorithm is a heuristic random search algorithm based on the evolution differences among different individuals, and it can exploit local regions within a neighborhood of an individual. To fully utilize characteristics of both algorithms, this paper for the first time proposes a hybrid algorithm based on HGS and DE. Furthermore, DE is also enhanced by a new ranking-based mechanism to generate more promising solutions. Experimental results on IEEE CEC2017 benchmark functions indicate the effectiveness of the hybrid algorithm, namely DEHGS, in comparison with some other state-of-the-art algorithms.

With the deregulation of the electricity market, the scale of market transactions and the number of access market members will grow explosively in China. In order to optimize the load of the entire server cluster in high concurrent access electricity market technical support system, a dynamic weight load balancing algorithm is proposed based on adaptive differential evolution (ADE) algorithm. The proposed algorithm considers the load balancing of data links as well as the service capability balancing of the server cluster. Firstly architecture of two-level electricity market technical support system is discussed. Then the principle of the ADE algorithm is discussed, and the improved dynamic load balancing is proposed with the ADE Algorithm, in which the server performance and link load balance parameters are dynamically evaluated and optimized. Finally, simulation and experimental are carried out with Nginx. The results show that the proposed dynamic weight load balancing algorithm has good performance of high concurrency, which can improve the system's maximum throughput and reduce response time of the entire cluster.

In order to further improve the performance of biogeography-Based optimization algorithm (BBO), an improved differential evolution biogeography optimization algorithm is proposed. By combining the search ability of differential evolution algorithm with the utilization of biogeography optimization algorithm, the elite retention mechanism is adopted to retain the individuals with high fitness, and the inertia weight strategy is introduced to adjust the proportion of mutation operation in the hybrid migration operation to improve the global search ability of the algorithm, and then the small probability perturbation is added to prevent the algorithm from changing with the iteration It is found that the optimal solution is local. Finally, six test functions are used for experiments, and the results show that the improved algorithm is better than other algorithms in optimization results, convergence speed and stability.

Multi-Robot system can be applied to search and rescue (SAR) tasks in complex environment. This paper focuses on resource scheduling and task allocation of cooperative multi-robot system for SAR (RSTA-RSSAR). RSTA-RSSAR is an extention of MSRCPSP. In RSTA-RSSAR, skills of multi-robot system considered from both depth and breadth. And resouece transfer time between tasks are considered. Meanwhile, Processing time of activities in RSTA-RSSAR changes with skill ability resouces provided. In order to solve the problem effectively, a genetic differential evolution algorithm (PS-GDEA) is proposed. An encoding, decoding and population initialization methods is designed according to the characteristics of RSTA-RSSAR. In this algorithm, a multi-attribute choosing method based on Shannon entropy is proposed. Meanwhile, a genetic algorithm with new selection, crossover and mutation operators is used to generate offspring and a priority based differential evolution strategy is used to evolute parents. The effectiveness of the algorithm is obtained by experiments on test cases. Compared with the classical genetic algorithm for msrcpsp, the proposed algorithm has better performance.

This paper aims to obtain optimal directional overcurrent relay (DOCRs) settings, which is a common problem in power system protection. This study was conducted in the industrial plant that used ring main distribution topology. Unlike previous research, this paper considers the impact of generator operation schemes. In addition, this paper also deals with the condition where one of the feeders is open on the ring main distribution system and change its topology to a radial network. Both of the conditions above have impact on DOCRs setting. The differential evolution (DE) algorithm is used to obtain the optimal time dial setting (TDS) for the relays connected in the network. Two variant mutation strategies of the DE algorithm have been implemented and compared respectively to fill the gap of previous study. The difference of operating time between DE/best/1 and PSO on all cases is above 25% while DE/best/1 and DE/rand-to-best/l is above 8.7%. This study concludes that the best mutation strategies method for solving DOCRs setting is DE/best/1. In addition, DE/best/1 has been compared with particle swarm optimization (PSO), the results concludes the DE/best/1 more optimized than PSO.

Challenges of managing and controlling wireless sensor networks (WSNs) includes efficient routing, increasing sensor node lifetime by conserving energy consumption, and base stations fault-tolerance. These challenges can be solved by help of Minimum Dominating Set (MDS) algorithms. By applying MDS to WSNs, only base stations (dominating sensors) take the burden of the communication instead of all sensors, and consequently saving the communication bandwidth, energy consumption, and increase the network lifetime. In addition to that, having multiple MDS improve base stations fault tolerance by scheduling different MDS periodically. In this research, a new Differential Evolutionary (DE) is composed to solve the general MDS problem. First, this problem is formulated as a binary optimization problem. Then, new evolutionary operations are introduced to fit the considered problem and wireless sensor network applications. Beside these evolutionary operations, other search enhancement operations are invoked to improve the search diversification and intensification process in addition to enabling the search process to find several distinct solutions. Those solutions can help in scheduling the network control nodes in order to increase its lifetime. Several experimental simulations over benchmark networks are carried out to test the efficiency of the proposed method. The results demonstrate the quality of the proposed method for obtaining minimum dominating sets for the considered test networks.

The increasing interest, in the last two decades, of evolutionary computation community in the combination of quantum computing and evolutionary computing, has led to the definition of a novel class of quantum-inspired evolutionary algorithms which exploit the concepts of quantum bits and quantum gates with the aim of improving the efficiency of current optimization methods. In this paper, a new method which hybridizes differential evolution with quantum computing is proposed. The method consists of two phases. In the first phase, it performs a new quantum-inspired evolutionary method until it does not get stuck into a local optimum, then, in the second phase, it runs differential evolution by using as initial population that obtained at the end of the first phase. Experiments on classical benchmark functions show that the hybridization outperforms standard methods by sensibly improving the fitness value and speeding up the convergence process.

The applications of parallel robots in various fields are increasing. In this paper, a research on the design optimization of a 6-PSS parallel manipulator for motion simulating is presented. The optimization problem of maximizing the workspace is formulated according to geometric and kinematic analysis. In particular, the workspace is a totally flexible workspace with 6 degrees-of-freedom (DOF). To maximize the workspace of the 6-PSS parallel robot, we investigated the differential evolution algorithm (DE), and the genetic algorithm (GA) is also adopted as a comparison. Computational results show DE algorithm outperforms GA algorithm at about 48.3% in optimal solution providing and 9.3% in time consuming.

The multi-layer kernel extreme learning machine (MKELM) is an improved multi-layer extreme learning machine(ML-ELM) based on kernel method, which has the advantages of no need to manually adjust the number of nodes in each hidden layer, no need to randomly initialize weights and offsets. However, due to the introduction of kernel method, it is necessary to select appropriate kernel parameters to obtain a model with better generalization performance, and manually adjust the parameters in kernel method according to prior knowledge, which is undoubtedly a very time-consuming process. In this paper, a self-adaptive differential evolution multi-layer kernel extreme learning machine (SaDE-MKELM) is proposed. The main idea of this algorithm is to use the self-adaptive differential evolution(SaDE) to automatically optimize the parameters in MKELM, take the parameter vector composed of the kernel parameter σ and penalty factor c of MKELM as an individual in the differential evolution algorithm, and build a MKELM model for each individual. The average root mean square error obtained by 10-fold cross-validation is used as the fitness of the corresponding individual, and the individuals in the population are selected by evolution within a certain optimization range and iteration times, and finally the optimal parameter combination suitable for the current model is obtained. The proposed algorithm is tested on several UCI simulation datasets. The experimental results show that the classification prediction accuracy of SaDE-MKELM is improved by at least 2% compared with MK-ELM and SaDE-ELM, and it has better stability.

The advent of Internet of Things (IoT) has bring a new era in communication technology by expanding the current inter-networking services and enabling the machine-to-machine communication. IoT massive deployments will create the problem of optimal power allocation. The objective of the optimization problem is to obtain a feasible solution that minimizes the total power consumption of the WSN, when the error probability at the fusion center meets certain criteria. This work studies the optimization of a wireless sensor network (WNS) at higher dimensions by focusing to the power allocation of decentralized detection. More specifically, we apply and compare four algorithms designed to tackle Large scale global optimization (LGSO) problems. These are the memetic linear population size reduction and semi-parameter adaptation (MLSHADE-SPA), the contribution-based cooperative coevolution recursive differential grouping (CBCC-RDG3), the differential grouping with spectral clustering-differential evolution cooperative coevolution (DGSC-DECC), and the enhanced adaptive differential evolution (EADE). To the best of the authors knowledge, this is the first time that LGSO algorithms are applied to the optimal power allocation problem in IoT networks. We evaluate the algorithms performance in several different cases by applying them in cases with 300, 600 and 800 dimensions.

The uncertain nature of renewable sources and load demand makes modern power systems highly complex and causes the frequency fluctuations. This research presents a maiden application of an Improved Sunflower Optimization Algorithm (ISFO) technique tuned Adaptive Type-2 Fuzzy PID (AT2FPID) structurefor frequency regulation of hybrid distributed power systems integrated with renewable sources. Initially, PID controller is considered and the dominance of ISFO over Sunflower Optimization Algorithm (SFO), Genetic Algorithm (GA) and Differential Evolution (DE) has been established. Performance comparison is carried out by assessing overshoots, undershoots and various integral errors due to Step Load Perturbations (SLPs) in each area. In the next stage, AT2FPID controller is considered and its supremacy to control the system frequency is demonstrated by comparing with Type-2 Fuzzy PID (T2FPID), Type-1 Fuzzy PID (T1FPID) and PID controllers.

Ensembles have been used in the evolutionary computation literature to evolve several populations in an independent manner, using different search approaches. Moreover, each population’s parents compete with their offspring and the other population’s offspring to improve diversity. It has been shown that ensemble algorithms improve the performance of the techniques embedded within them, when considered independently. Furthermore, scalarizing functions have been successfully used in decomposition-based and some indicator-based Multi-objective Evolutionary Algorithms (MOEAs). However, it has been shown that the performance of scalarizing function tends to be tied to the geometrical shape of the Pareto front. In this work, we propose a new ensemble algorithm that adopts different scalarizing functions and weight vectors using Hungarian Differential Evolution as the baseline multi-objective optimizer. Our experimental study shows that our proposed approach outperforms the original HDE, and it is competitive with respect to modern MOEAs.

The performance of differential evolution (DE) algorithm depends critically on the setting of mutation factor F and crossover rate CR. In this paper, a multi-factor ranking based parameter adaptation scheme is proposed to properly set the value of F and CR. The proposed adaptation scheme includes a parameter storage and distribution mechanism. The parameter storage mechanism is used to produce a suitable value by considering the multi-factor ranking scheme, which can help the algorithm enhance its exploration ability in the early stage. The parameter distribution mechanism is a layered parameter calculation strategy, in which the individuals generate new parameters based on multi-factor ranking properties. This mechanism is helpful to improve the exploitation capability of DE. The performance of the proposed method has been evaluated on CEC2014 and CEC2015 test suites and compared with related methods. The results show our method could achieve good performance and outperform related methods.

With the increase of scenes where Neural Networks are used as a classifier, the expectation of the classifying accuracy for the network has risen. To improve classifying accuracy, Differential Evolution (DE) has been applied as an optimization method for Neural Networks. Compared to other DE methods, Differential Evolution with an Individual-Dependent Mechanism (IDE) takes in account of the differences between the fitness value of individuals. As a result, IDE has better results in terms of accuracy. Therefore in this paper, a Neural Network optimizer using IDE is proposed for further Neural Network improvement. Moreover, while most optimizers would use the loss function as the fitness value in DE, a method using accuracy is proposed because DE does not require the activation function to be differentiable. Experiments using the proposed framework has been conducted on classification problems, and comparative studies with other self-adaptive mutations and traditional optimizing methods have been performed. Experimental results showed that the proposed method outperformed other conventional methods in terms of accuracy.