ACEPSO: A multiple adaptive co-evolved particle swarm optimization for solving engineering problems

Gang Hu

doi:10.1016/j.aei.2024.102516

2024-08-01

PRE

DOI：10.1016/j.aei.2024.102516

原文链接

原文求助

摘要

En 中文

Particle swarm optimization (PSO) is one of the most classical metaheuristic algorithms that has gained significant attention since its inception. It has some inherent advantages, such as easy implementation, rapid convergence, low computational complexity and so on. However, the drawbacks of being prone to local optimization and insufficient diversity cannot be ignored. Therefore, a new multiple adaptive co-evolved particle swarm algorithm (ACEPSO) with adaptive population grouping strategy, pros-cons coevolution mechanism, new co-evolved mechanism and adaptive mutation strategy is proposed in this paper. Firstly, ACEPSO partitions the overall population into two distinct subpopulations: elite population and common population. The size of the subpopulations undergoes variations at different stages. Secondly, the introduced pros-cons coevolution mechanism effectively improves the exploration ability of PSO. Meanwhile, a new co-evolved mechanism is proposed here aiming to enhance population diversity and balance the exploration and exploitation ability. This mechanism can better transfer information between individuals and promote effective collaboration. Finally, an adaptive mutation strategy is introduced. It improves the population diversity and prevents the algorithm from falling into local optimality productively. To validate the outstanding performance of ACEPSO, this paper compares it with various state-of-the-art metaheuristic algorithms as well as their variants on CEC2017 and CEC2022 test sets. The results exhibit that ACEPSO has a standout comprehensive performance. In addition, ACEPSO is utilized to tackle a set of twelve engineering optimization problems as well as 2D robot path planning problems. On all these complex optimisation problems, ACEPSO obtains the relatively best results. All the above results manifest that ACEPSO has great advantages and competitiveness in solving some of the optimization problems.

Keyword:

Particle swarm algorithm

Pros -cons coevolution mechanism

Co -evolved mechanism

Engineering optimization problems

2D robot path planning problems