Traditional multi-modal multi-objective evolutionary algorithms (MMEAs) generally do not fully exploit the dynamic changes of population distributions across generations to enhance algorithmic exploitation ability. Therefore, this paper proposes dual-space co-evolution with temporal information learning for multi-modal multi-objective optimization evolutionary algorithm (Ds-TILEA). Ds-TILEA proposes a temporal information learning framework that transforms historical population information into time series and utilizes a deep neural network to forecast population distributions. Unlike conventional time-series-assisted evolutionary methods that mainly use historical information to predict external environments, scalar indicators, or auxiliary search states, Ds-TILEA treats generation-indexed population distributions as endogenous multivariate temporal data and directly uses the predicted decision-space positions for offspring generation. Specifically, the predictive model is developed to capture the evolutionary patterns of adjacent generations’ population distributions, thereby facilitating the generation of potentially high-quality offspring. Furthermore, achieving a balance between diversity and convergence of Pareto sets is a critical challenge in multi-modal multi-objective optimization problems (MMOPs). Accordingly, this study designs a dual-space updating strategy, where in the diversity archive incorporates dual crowding-distance calculations in both decision and objective spaces to improve solution diversity while maintaining convergence. To validate the performance of Ds-TILEA, experiments comparing it with six competing algorithms are conducted on two standard benchmark suites. Furthermore, seven algorithms were applied to a feature selection task aimed at minimizing dataset dimensionality and classification error rate. The experimental results demonstrate that Ds-TILEA exhibits superior performance in both convergence and diversity. Specifically, among the 12 IDMP test problems, Ds-TILEA achieved 6, 4, and 8 best optimal values in IGD, IGDX, and HV, respectively, with statistical significance confirmed by the Wilcoxon rank-sum test. Ds-TILEA demonstrates significant advantages in MMOPs, especially in efficiently balancing diversity and convergence by integration of temporal information to generate offspring.

When solving constrained multi-objective optimization problems (CMOPs), traditional methods often utilize a holistic constraint violation, which obscures the unique effect of individual constraints. Recently, methods based on constraint decomposition or prioritization have been proposed to consider constraints individually. However, those methods typically rely on pre-defined evaluation rules, which lack the ability to dynamically exploit the effect of an independent constraint for enhancing the exploration of the constrained Pareto front (CPF). To address this limitation, a Dynamically Competitive Constraint Handling algorithm (DCCHT) is proposed for Constrained Multi-objective Optimization. Specifically, DCCHT decomposes a CMOP with K constraints into K+1 distinct optimization pools, where one optimization pool considers all constraints and another K optimization pools respectively involve each constraint among the overall K constraints. The information exchange among distinct optimization pools is facilitated by a proposed bidirectional knowledge transfer strategy, which shares valuable landscape information and accelerates the overall convergence to the global feasible region. The ongoing progress of each optimization pool is dynamically controlled based on the competitiveness of the constraint. Computational resources are adaptively allocated to the optimization pool by a feedback-driven resource allocation method. More computational resources will be allocated to the optimization pool in which the considered constraint is more competitive with the exploration of the true CPF. Extensive experiments on three benchmark suites and various real-world engineering problems demonstrate the significantly better performance of DCCHT against some state-of-the-art algorithms.

The construction of structural equation models (SEMs) is fundamentally challenged by the latent nature of the constructs, which often leads to ineffective model specification and computational inefficiency in many-objective optimization scenarios. Conventional many-objective evolutionary algorithms frequently exhibit instability and a tendency to produce lots of infeasible solutions when applied to many-objective SEMs. To overcome these limitations, this paper proposes a dynamic regularization-based evolutionary learning algorithm (DR-ELA). The proposed DR-ELA incorporates a novel dual-matrix fusion model for robust latent structure discovery, coupled with an innovative tri-phase dynamic regularization mechanism. This mechanism automates sparsity control through adaptive germination, development, and maturation phases. Experimental results on many-objective SEM test instances demonstrate that the proposed DR-ELA performs significantly better than several many-objective evolutionary algorithms. Furthermore, experimental results show that DR-ELA can effectively balance theoretical fidelity and exploratory through the adaptive regularization mechanism, achieving superior performance in both solution quality and computational efficiency.

The core challenge for multimodal multi-objective problem (MMOP) resolution lies in maintaining synergistic interactions between convergence and diversity. However, the existing algorithms usually consider convergence-first, which neglect to consider both diversity and convergence into account during the evolutionary process. Likewise, the optimization methods tend to gravitate toward locally optimal regions rapidly, leading to lose diversity for the local PS. This paper proposes a Deep Reinforcement Learning-guided multimodal multi-objective evolutionary algorithm with a serial-parallel mechanism (DRLMMEA) to investigate the impact of different operator selection on the performance of MMEAs, which greatly helps to balance the convergence and diversity. DRLMMEA utilizes Q-Network to select the operator with the highest reward to enhance the population's search ability. An improved sorting method (ISM) based on neighborhood dominance updates the population by sorting individuals according to their convergence quality, thereby enhancing convergence performance in the objective space. Moreover, this study proposes a series-parallel mechanism, a series structure enhances the diversity in the decision space, while the parallel structure reduces the computational burden of the algorithm evidently. The proposed Deep Reinforcement Learning-assisted operator selection mechanism, which enables effective balance between diversity and convergence, and an improved crowding distance approach that enhances convergence performance. DRLMMEA undergoes comprehensive testing against 6 contemporary approaches using MMF and IDMP benchmark problems, achieving supremacy in 4 principal performance metrics according to experimental findings. The multimodal gearbox parameter optimization is addressed using the proposed DRLMMEA, which demonstrates superior performance against 6 algorithms in comparative evaluations. It has demonstrated a significant role in solving the MMOPs with the imbalance between convergence and diversity.

As the number of objectives of the optimization problem increases, traditional many-objective evolutionary optimization algorithms face great challenges when balancing convergence and diversity. This is due to the low selection pressure towards the Pareto front and the high proportion of non-dominated and dominance-resistant solutions in the population. Moreover, existing indicator-based many-objective algorithms often employ a single indicator to evaluate solutions, which makes the quality of solutions largely determined by how well the indicator is designed. To address the aforementioned issues, this paper employs a cooperative coevolutionary approach by dividing the decision variables into convergence-related and diversity-related decision variables, each optimized in its respective subpopulation with distinct evaluation indicators. The two subpopulations coevolve cooperatively to better balance convergence and diversity. The proposed algorithm is compared with eight MaOEAs on three benchmark problems of DTLZ, WFG, and MaF. Experimental results demonstrate that the proposed algorithm exhibits competitive performance in solving many-objective optimization problems.

Constraint multiobjective algorithms are the most widely applied direction in intelligent optimization, with excellent research value. Currently, most multiobjective multi-constraints algorithms are designed based on the relationship between feasible and infeasible solutions, but they ignore the complex associations between constraints. At the same time, the most significant difficulty in constraint problems lies in the irregularity of constrained Pareto front (CPF) and the lack of a powerful strategy for exploration. The paper proposes a Pareto front searching based on reinforcement learning (PFRL) for multi-constraints multiobjective optimization problems. The algorithm employs reinforcement learning to guide the evolution process through interaction with the environment and adaptively learns the shape and characteristics of CPF to cover the structure of CPF effectively. The environment and CPF information gained by reinforcement learning are utilized for CPF translation and extension to deal with various irregular feasible regions. In addition, the paper also designs a constraint priority evaluation mechanism based on the correlation distance (CD) metric to process constraint relationships. It allows the algorithm to effectively cross over Pareto front (PF) of a single constraint that is unrelated to CPF, improving algorithm efficiency. The introduced algorithm implemented the above strategy using only one population. The effectiveness of the introduced algorithm was verified and compared with nine state-of-the-art algorithms and four real-world constrained multiobjective optimization problems (CMOPs). Experimental results show that the algorithm provides a low-resource and efficient method for solving CMOPs.

Smart greenhouse is a modern agricultural facility that integrates smart control systems to regulate the plant growth environment through advanced intelligent technology and devices. In recent years, smart greenhouses have received widespread attention and have been applied in agriculture. Due to their high energy demands and costs, current smart greenhouses are often impractical for applications with limited resources. Nevertheless, small-scale smart greenhouse with low-power microprocessor, are more suitable for homes, offices, and other fields. Therefore, this paper proposes a micro dynamic multi-objective evolutionary algorithm ( mu DMOEA) for small-scale smart greenhouse with low-power microprocessor, which applies chaotic mapping to select dynamic response strategies based on the fitness of dominant relationships and k-nearest neighbor environmental selection. mu DMOEA performs well in the simulation of small-scale smart greenhouses. It not only outperforms SGEA, DNSGA-II, and RVCP in IGD indicator but also plays a good role in adjusting environmental parameters. It demonstrates the feasibility and effectiveness of micro dynamic multi-objective optimization on small-scale smart greenhouse with low-power microprocessor.

Particle swarm optimization (PSO) algorithms have been successfully applied to all kinds of optimization problems. However, the standard PSO algorithm can easily fall into local optimal regions when solving complex problems. It is difficult for the particles to get out of these regions, resulting in the inability to obtain the global optimal solution. This paper proposes a dynamic population synergy particle swarm optimization algorithm (DPSPSO) to solve the above problem. The DPSPSO algorithm dynamically divides the population into three sub-populations based on the function-adapted value ordering strategy. Each sub-population has a different task depending on its potential. Moreover, the algorithm employs an integrated learning strategy. This strategy aims to fully utilize the practical information provided by the particles to prevent premature convergence of the particles. Finally, this paper compares the proposed algorithm with five state-of-the-art PSO variants on the CEC2022 test function to investigate the algorithm's effectiveness. The experimental results show that DPSPSO can achieve competitive performance compared to the other PSO variants.

With the advancement of the manufacturing industry, the investigation of the shop floor scheduling problem has gained increasing importance. The Job shop Scheduling Problem (JSP), as a fundamental scheduling problem, holds considerable theoretical research value. However, finding a satisfactory solution within a given time is difficult due to the NP-hard nature of the JSP. A co-operative-guided ant colony optimization algorithm with knowledge learning (namely KLCACO) is proposed to address this difficulty. This algorithm integrates a data-based swarm intelligence optimization algorithm with model-based JSP schedule knowledge. A solution construction scheme based on scheduling knowledge learning is proposed for KLCACO. The problem model and algorithm data are fused by merging scheduling and planning knowledge with individual scheme construction to enhance the quality of the generated individual solutions. A pheromone guidance mechanism, which is based on a collaborative machine strategy, is used to simplify information learning and the problem space by collaborating with different machine processing orders. Additionally, the KLCACO algorithm utilizes the classical neighborhood structure to optimize the solution, expanding the search space of the algorithm and accelerating its convergence. The KLCACO algorithm is compared with other high-performance intelligent optimization algorithms on four public benchmark datasets, comprising 48 benchmark test cases in total. The effectiveness of the proposed algorithm in addressing JSPs is validated, demonstrating the feasibility of the KLCACO algorithm for knowledge and data fusion in complex combinatorial optimization problems.

Download This Paper Open PDF in Browser Add Paper to My Library Share: Permalink Using these links will ensure access to this page indefinitely Copy URL Copy DOI

Adaptive path planning and optimization for robots are a persistent challenge due to the lack of a cognitive map of the complex and dynamic environments. This paper presents a cognitive-map-based method for dynamic robot path planning by a formal maze model for workplace layouts representation. The maze model and global path trees provide a cognitive map for the robot to aware the environment. A theory of dynamic Paths Finding and Optimization (PFO) is developed. Powered by the PFO algorithm, a robot is able to autonomously explore the optimal path in a complex workplace by a single sight-read of its layout. The maze based PFO methodology provides robot for an efficient real time path cognition and optimization algorithm for dealing with dynamic workplaces. A set of experiments demonstrates the efficiency of the method, which extends traditional stepwise robot vision technologies to cognitive-map-driven global path optimization.

Due to the problems of low accuracy and increasing control parameters in the existing parameter adaptive methods of differential evolution (DE) algorithm, in this paper a muta-tion operator selector and a parameter selector are proposed through Fitness Landscape (FL) analysing. At first, the performance differences of the two categories of mutation oper-ators named DE/best/1 and DE/current -to -rand/1 were analyzed on many test prob-lems. Secondly, the relationship between the FL and mutation operator is founded by using ensemble learning and decision tree, and achieved a classifier named mutation oper-ator selector. Thirdly, the relationship between the FL and algorithm parameters is founded by using a neural network, and then a classifier named parameter selector is achieved. Finally, the improved DE algorithm equip with the two selectors is tested on the CEC2017 benchmark set. The results show that the proposed improved DE algorithm is outperforms both the basis DE algorithm and other three state-of-the-arts algorithms. (c) 2022 Published by Elsevier Inc.

The particle swarm optimization (PSO) algorithm has the characteristics of fewer parameters, fast convergence speed and easy implementation. However, the algorithm has the problem of low accuracy and tends to fall into local optimization. In this paper, a particle swarm optimization algorithm with dual population adaptive mutation (DPAMPSO) is proposed. In this algorithm, the whole population is divided into a general subpopulation and an elite subpopulation. In order to balance the global exploration and local exploitation ability of particles, a chaos-based inertia weight is proposed in this paper. Among them, the ordinary subpopulation uses an adaptive mutation strategy. By setting the probability parameter to make the common subpopulation randomly choose to mutate in the direction of the superior particle or still keep its own direction. The elite subpopulation then uses an exemplar learning strategy. The historical optimal and global optimal individuals of the particles are crossed to generate examples, thus guiding the particles to perform a local search for the region of possible solutions. To verify the effectiveness of the algorithm, the DPAMPSO algorithm is tested on ten benchmark problem. It is also compared with other recent variants of particle swarm optimization algorithms. The experimental results show that the DPAMPSO algorithm is superior to other variations of PSO in terms of solving accuracy and convergence speed.

The particle swarm optimization algorithm has been widely utilized to address a wide range of different engineering optimization problems due to its few parameters and simple structure. The conventional particle swarm optimization algorithm takes in information from two sources: Global optimal particle and individual optimal particle. However, learning from these two sources alone is inefficient to solve complex high-dimensional problems. Therefore, this paper proposes a multi-population cooperative particle swarm optimization (COVPSO) algorithm with covariance guidance strategy, through the use of a covariance guidance strategy to guide population evolution direction. In COVPSO, the population is divided based on the Euclidean distance from the particle to the global optimal particle, and the population is divided into the elite group, exploratory group, and inferior group. As a result of grouping the population and adopting different strategies, the elite group has good exploitation ability, while the exploration group has good exploration ability, and the inferior groups by introducing a differential mutation operator to improve global exploration ability. Therefore, the population has a good balance between exploration and exploitation. This study utilizes ten benchmark functions and five PSO variants broadly used in the literature to verify the merits of COVPSO to demonstrate its efficiency. The findings of the experiments show that COVPSO has a faster convergence rate and a more precise solution.

The 0–1 multidimensional knapsack problem is an NP-hard combinatorial optimization problem, which is widely used in life. The existing heuristic algorithm suffers from the problem of high time consumption or low solution accuracy when faced with a large-scale and high-dimensional knapsack problem. This paper proposes an adaptive search algorithm with scatter and tabu strategy. The algorithm integrates adaptive scatter search and solution-based tabu search into the evolutionary framework. In the search process, adaptive reference set strategies, repeated flip mutations and the neighborhood structure that limits the step size change are used. In order to verify the performance of the algorithm, the algorithm was tested with 60 benchmarks, compared with other algorithms in the international standard library. Experimental results show that the proposed algorithm is competitive with other algorithms.

Particle swarm optimization has been widely utilized to tackle various real optimization problems as an effective and simple optimization approach. However, the phenomenon of premature convergence has always existed. To ameliorate the drawback, a self-learning particle swarm optimization algorithm with multi-strategy selection (MSLSPSO) is proposed in this study. In MSLSPSO, an elite particles guidance strategy is proposed. This strategy makes elite particles participate in the search process to strengthen the guidance of the population. A Lévy Flight perturbation strategy is designed, which utilizes the random walk characteristic of Lévy Flight to perturb the historical and global optimal of particles, increasing the diversity of the population. In addition, a fitness distance correlation self-learning mechanism is proposed, which can self-learning according to the characteristics of the population in the search process. A set of test functions are utilized for experimental analysis and compared with four well-known PSO variants. Experimental results show that the MSLSPSO algorithm has efficient robustness and competitive solutions can be obtained.

Particle swarm optimization (PSO) has attracted wide attention in the recent decade. Although PSO is an efficient and simple evolutionary algorithm and has been successfully applied to solve optimization problems in many real-world fields, premature maturation and poor local search capability remain two critical issues for PSO. Therefore, to alleviate these disadvantages, a dynamic population cooperative particle swarm optimization for global optimization problems (DPCPSO) is proposed. Firstly, to enhance local search capability, an elite neighborhood learning strategy is constructed by leveraging information from elite particles. Meanwhile, to make the particle easily jump out of the local optimum, a crossover-mutation mechanism is utilized. Finally, a dynamic population partitioning mechanism is designed to balance exploration and exploitation capabilities. 16 classic benchmark functions and 1 real-world optimization problem are used to test the proposed algorithm against with 6 typical PSO algorithms. The experimental results show that DPCPSO is statistically and significantly better than the compared algorithms for most of the test problems. Moreover, the convergence speed and convergence accuracy of DPCPSO are also significantly improved. Therefore, the algorithm is highly competitive in solving global optimization problems.

Citrus is the largest fruit production in the world. Owing to the damage by various pest diseases, the production of citrus is reduced and the quality is getting worse and worse every year. The recognition and control of the citrus diseases are very important. By now the main measures we take to control them is sowing pesticides, which is not good for the environment and do harm to the soil greatly. The technology of image identification can recognize what kind of citrus disease they have with high efficiency and low cost, which is also environmentally friendly and is not limited by time and space. It is our top priority to apply it to recognize and prevent the disease from citrus. In order to detect citrus pest disease and control them automatically, we studied the pests and traits of citrus leaves and their multi-fractal characteristics and methods for figuring pests and diseases, and created a model for detecting leaf images of citrus. We use Keras and Tensorflow to build the model. To reduce recognition loss and improve accuracy, we put the citrus photos into the model and train it persistently. After examining, the recognition accuracy of citrus greening disease of 120 images can reach 96%. The experimental result shows that the model can recognize citrus diseases with high accuracy and robustness.

Rail guide vehicle (RGV) dynamic scheduling problems have attracted increasing attention in recent years, which determines a great impact on the working efficiency of the entire scheduling system. However, the relative intelligent optimization study of RGV dynamic scheduling problems are insufficient scheduling of different working components in the previous works, it is easy to appear idle waiting, resulting in reduced operating efficiency during operation. Analysis of the fitness landscape is essential to understand the behavior of evolutionary algorithms for solving dynamic optimization problems in the evolutionary dynamics of biological evolution. With the continuous advancement of evolutionary algorithm optimization, the fitness landscape can present more abundant feature information around the fitness value, including autocorrelation, fitness distance correlation, landscape walks, local optima, and landscape roughness. This paper proposes a new distance landscape strategy for the RGV dynamic scheduling problems. The combination of the fitness landscape and dynamic search strategy are established according to the operating principle of the RGV system. In order to solve the RGV dynamic scheduling problem more effectively, experiments are conducted based on the type of computer numerical controller (CNC) with one procedure programming model in solving the RGV dynamic scheduling problems. The experiment results reveal that this new distance landscape strategy can provide promising results and solve the considered RGV dynamic scheduling problems effectively.

Optimization problems widely exist in scientific research and engineering practice, which have been one of the research hotshots and difficulties in intelligent computing. The single swarm intelligence optimization algorithms often show such defects as searching stagnation, low accuracy of convergence, part optimum and poor generalization ability when facing the increasingly sophisticated optimization problems. In the study of multiple population, the choice of evolution strategy often has great influence on the performance of the algorithm, and this paper puts forward a kind of dual-population evolutionary algorithm adapting to complementary evolutionary strategy (DPCEDT) based on the study of differential evolution algorithm, teaching and learning-based optimization algorithm. The simulation results show that the algorithm performs better than the TLBO-DE, HDT and DPDT and some other algorithms do in most test functions. It suggests that the complementary evolutionary strategies are more advantageous than other evolutionary strategies in dual-population evolutionary algorithms.