一种自适应离散粒子群算法及其应用研究

A Self-Adaptive Discrete Particle Swarm Optimization Algorithm

本文提出了一种改进的离散粒子群算法.为了克服算法的早熟收敛问题,引入了一个排斥过程用于增加群体的多样性,提出了一种控制群体多样性的准则,实现了算法运行过程中吸引和排斥过程的动态自适应切换.为了提高算法的收敛速度,提出了一种惯性权重动态变化策略,在算法执行的不同阶段,使惯性权重随迭代次数动态自适应变化.试验中发现,引入局部搜索技术后,算法的性能会进一步提高.最后将此算法用于解决TSP问题及车间调度问题并与其他相关算法进行了比较,实验结果表明,收敛速度快,稳定性强.

A self-adaptive discrete particle swarm algorithm is proposed.In order to overcome the premature convergence of the algorithm,a repulsive process is introduced to increase the swarm diversity and a metric to measure the swarm diversity is also designed.The attractive and repulsive processes can adaptively change during running.To speed up convergence,a strategy used to control the inertia weight is advanced which changes dynamically with the iterations during different running phrase of the algorithm.Moreover,algorithm performance can be enhanced further if local search strategies are combined.Finally,the proposed algorithm is used to solve the TSP and FSSP problems and compared with other related algorithms.The experiment results showed its superiority.