一种求解大规模问题的自学习协同粒子群算法

AN SELF-LEARNING COOPERATIVE PARTICLE SWARM OPTIMIZATION ALGORITHM FOR SOLVING LARGE SCALE PROBLEMS

随着工程技术要求的提高,许多实际优化问题从低维问题发展成高维的大规模优化问题,自然计算算法在面对该类问题时容易陷入局部最优,而协同粒子群算法是解决大规模优化问题的重要手段之一。本文将子种群划分自学习策略和惯性权重自适应策略引入到协同粒子群算法中,增强了算法的自学习能力,提高了算法的全局寻优能力。实验结果表明,所提算法的性能超过了传统协同粒子群等算法,具有求解大规模问题的较大潜力。

With the improvement of engineering requirements, many practical optimization problems have been developed from low dimensional problems to large-scale optimization problems. The natural computing algorithm is prone to fall into the local optimal in the face of this kind of problem. Cooperative particle swarm optimization（CPSO） is one of the important means to solve large-scale optimization problems. In this paper, sub population partition self-learning strategy and inertia weight self-adaptive strategy are introduced into cooperative particle swarm optimization algorithm, which enhances the self-learning ability of the algorithm and improves the global optimization ability of the algorithm. The experimental results show that the performance of the proposed algorithm exceeds the traditional cooperative particle swarm optimization algorithms, and has great potential for solving large-scale problems.