Search

When tackling imbalanced constrained multi-objective optimization problems (CMOPs) with simultaneous convergence-hard and diversity-hard constraints, a critical issue is to balance the diversity and convergence of populations. To address this issue, this paper proposes a hybrid algorithm which combines an improved epsilon constraint-handling method (IEpsilon) with a multi-objective to multi-objective (M2M) decomposition approach, namely M2M-IEpsilon. The M2M decomposition mechanism in M2M-IEpislon has the capability to deal with imbalanced objectives. The IEpsilon constraint-handling method can prevent populations falling into large infeasible regions, thus improves the convergence performance of the proposed algorithm. To verify the performance of the proposed M2M-IEpsilon, a series of imbalanced CMOPs with simultaneous convergence-hard and diversity-hard constraints, namely ICD-CMOPs, is designed by using the DAS-CMOPs framework. Six state-of-the-art constrained multi-objective evolutionary algorithms (CMOEAs), including CM2M, CM2M2, NSGA-II-CDP, MOEA/D-CDP, MOEA/D-IEpsilon and PPS-MOEA/D, are employed to compare with M2M-IEpsilon on the ICD-CMOPs. Through the analysis of experimental results, the proposed M2M-IEpsilon is superior to the other six algorithms in solving ICD-CMOPs, which illustrates the superiority of the proposed M2M-IEpsilon in dealing with ICD-CMOPs with simultaneous convergence-hard and diversity-hard constraints.

This paper gives a short review on mechatronic design automation (MDA) whose optimization method is mainly based on evolutionary computation techniques. The recent progress and research results of MDA are summarized systematically, and the challenges and future research directions in MDA are also discussed. The concept of MDA is introduced first, research results and potential challenges of MDA are analyzed. Then future research directions, focusing on constrained multiobjective optimization, surrogate-assisted constrained multi-objective optimization, and design automation by integrating constrained multi-objective evolutionary computation and knowledge extraction, are discussed. Finally, we suggest that MDA has great potential, and may be the next big technology wave after electronic design automation (EDA).

Multi-agent systems are applied to a variety of scenarios, in which target entrapment has become a primary research area in recent decades. In order to solve the problem of intelligent swarm behavior control, the hierarchical gene regulation network (H-GRN) is proposed. However, the networks in H-GRN rely solely on target information for behavioral control, and interaction with surrounding partners only involves avoiding physical collisions. To benefit from the cooperation with partners, we design a cooperation-based gene regulatory network (C-GRN) for target entrapment. Following the hierarchical gene regulatory network, we use the agent’s own sensor to get the companion information, and add information to the network by controlling changes in the corresponding protein concentration. In addition, a self-organizing obstacle avoidance control method is also proposed. A series of empirical evaluations index comparison show that C-GRN can cooperate with partners. The experimental results indicate that the total time to complete task and average thickness of the target’s encirclement is obviously optimized in a simulation experiment.

Collective task in swarm robots has been studied widely because of the ability limitation of a single robot. Collective tracking is an important ability for swarm, and many of previous tracking tasks are based on leader-follower model. Unfortunately, simple following behavior brings much tracking uncertainty in constrained environment and difficulty for a convergence tracking pattern. To address this issue, we propose a new model for tracking by combining tracking-based hierarchical gene regulatory network with leader-follower model named (TH-GRN) for swarm robots. The TH-GRN model simulates the process that proteins are generated and diffused to control swarm activities. The concentration diffusion forms a tracking pattern and guides swarm robots to designated pattern. In order to be adaptive to confined environment, some flexible strategies are devised and integrated into our proposed TH-GRN model to achieve better performance. Besides, the TH-GRN model is also used to generate dynamic and complex environment. In our experiments, we design three obstacle scenarios, i.e., fixed obstacles, mobile (dynamic) obstacles, and hybrid obstacles. We conduct some simulation to validate the effectiveness of tracking-based TH-GRN model, and the experiment results demonstrate the superiority of our model.

Choroidal neovascularization (CNV), which will cause deterioration of the vision, is characterized by the growth of abnormal blood vessels in the choroidal layer. Estimating the area of CNV is important for proper treatment and prognosis of the disease. As a noninvasive imaging modality, optical coherence tomography (OCT) has become an important modality for assisting the diagnosis. Due to the number of acquired OCT volumes increases, automating the OCT image analysis is becoming increasingly relevant. In this paper, we train a convolutional neural network (CNN) with the raw images to estimate the area of CNV directly. Experimental results show that the performance of such a simple way is very competitive with the segmentation based methods. To explain the reason why the CNN performs well, we try to find the function being approximated by the CNN. Thus, for each layer in the CNN, we propose using a surrogate model, which is desired to have the same input and output with the layer while its mathematical expression is explicit, to fit the function approximated by this layer. Genetic programming (GP), which can automatically evolve both the structure and the parameters of the mathematical model from the data, is employed to derive the model. Primary results show that using GP to derive the surrogate models is a potential way to find the function being approximated by the CNN.

In this paper, we discuss a discrete-time Geom\({^{[X]}}\)/G/1 queueing system with modified T vacation policy and startup time. We derive the generating functions and the mean values for the steady state system size and the waiting time, and also get those of the busy period, the vacation period and the vacation cycle by using embedded Markov chain. Finally, we determine the optimal \({(T^*,J^*)}\) to minimize the cost function with fixed cost elements by constructing a cost function.

This paper studies a continuous time queue system with second optional service where all the arriving customers demand the first “essential” service while only some of them demand the second “optional” service with probability \(\alpha \). The service time of the first essential service and the second optional service both are independent and arbitrarily random variables. Whenever a busy period is completed, the server takes a vacation. If there is at least one customer waiting at a vacation, the server immediately serves the customer. Otherwise, the server takes another vacation with probability p, or remains idle with probability \(1-p\). We give some performances analysis of this model. Finally, it gives some numerical examples to illustrate the effect of the probabilities \(\lambda \) and p on the mean system size, waiting time, the probabilities when the server is idle and is on a vacation.

The concrete delivery problem (CDP) is an NP-hard, real world combinatorial optimization problem. The CDP involves tightly interrelated routing and scheduling constraints that have to be satisfied by considering the tradeoff between production and distribution costs. Various exact and heuristic methods have been developed to address the CDP. However, due to the limitation of the exact methods for dealing with such a complex problem, (meta-)heuristics have been more popular. For this purpose, the present study proposes a hybrid algorithm combining simulated annealing (SA) with a time-slot heuristic (TH) for tackling the CDP. The TH is applied for generating new solutions through perturbation while simulated annealing is utilized to decide on whether to accept these solutions. The proposed algorithm, i.e. SA-TH, is compared to an existing CDP heuristic on a diverse set of CDP benchmarks. The computational results conducted through a series of experiments validate the efficiency and success of SA-TH.

The present paper proposes a Greedy based Pareto Local Search (GB-PLS) algorithm for the bi-objective robust airport gate assignment problem (bRAGAP). The bRAGAP requires to minimize the total passenger walking distance and the total robust cost of gate assignment. The robust cost is measured through our proposed evaluation function considering the impact of delay cost on the allocation of idle time. GB-PLS uses the Random and Greedy Move (RGM) as a neighborhood search operator to improve the convergence and diversity of the solutions. Two populations are maintained in GB-PLS: the external population (EP) stores the nondominated solutions and the starting population (SP) maintains all the starting solutions for Pareto local search (PLS). The PLS is applied to search the neighborhood of each solution in the SP and the generated solutions are used to update the EP. A number of extensive experiments has been conducted to validate the performance of GB-PLS over Pareto Simulated Annealing (PSA).

This paper proposes a decomposition-based multi-objective multi-factorial evolutionary algorithm (MFEA/D-M2M). The MFEA/D-M2M adopts the M2M approach to decompose multi-objective optimization problems into multiple constrained sub-problems for enhancing the diversity of population and convergence of sub-regions. An machine learning model augmented version is also been implemented, which utilized discriminative models for pre-selecting solutions. Experimental studies on nine multi-factorial optimization (MFO) problem sets are conducted. The experimental results demonstrated that MFEA/D-M2M outperforms the vanilla MFEA on six MFO benchmark problem sets and achieved comparable results on the other three problem sets with partial intersection of global optimal.

This paper presents a novel multiobjective constraint handling approach, named as MOEA/D-CDP-ID, to tackle constrained optimization problems. In the proposed method, two mechanisms, namely infeasibility driven (ID) and constrained-domination principle (CDP) are embedded into a prominent multiobjective evolutionary algorithm called MOEA/D. Constrained-domination principle defined a domination relation of two solutions in constraint handling problem. Infeasibility driven preserves a proportion of marginally infeasible solutions to join the searching process to evolve offspring. Such a strategy allows the algorithm to approach the constraint boundary from both the feasible and infeasible side of the search space, thus resulting in gaining a Pareto solution set with better distribution and convergence. The efficiency and effectiveness of the proposed approach are tested on several well-known benchmark test functions. In addition, the proposed MOEA/D-CDP-ID is applied to a real world application, namely design optimization of the two-stage planetary gear transmission system. Experimental results suggest that MOEA/D-CDP-ID can outperform other state-of-the-art algorithms for constrained multiobjective evolutionary optimization.

Open Architecture Product (OAP) offers public interfaces beyond the individual product. The interface can be shared by other products to enrich the product function and adaptability. Adaptable design (AD) meets OAP objectives for different requirements through modification or adaptation of product modules in the product lifecycle. The product adaptability is achieved by adaptable design, adaptable modules and platforms, and interfaces. This paper introduces a miniature electric vehicle with open architecture developed using AD. The electric vehicle consists of common platforms, customized modules, and user personal components. The vehicle developed can easily meet the individualization of users’ requirements and requirement changes in its lifecycle.