粒子群优化算法的边界变异策略比较研究

Comparative Study of Boundary Mutation Strategy for Particle Swarm Optimization Algorithm

为解决粒子群优化(PSO)算法中粒子越界和早熟收敛等问题,在比较国内外学者提出的边界变异策略基础上,提出一种新的边界变异策略——双重限制变异策略。针对粒子越界时速度和位置变异方向的不同情形,通过同时限制粒子的更新位置和更新速度,将粒子控制在搜索空间范围内。利用5种测试函数进行实验,结果表明,与其他4种边界变异策略相比,双重变异策略收敛速度快,在解决粒子越界问题上具有较好的效果。此外,通过实验测试显示粒子的最大速度和最大位置的比值与变异策略的好坏程度成反比,为边界变异策略的研究提供了一定依据。

To control particles to fly inside search space and deal with the problems of premature convergence of Particle Swarm Optimization（PSO）algorithm,based on the comparative study of boundary mutation strategy proposed by scholars at home and abroad,this paper proposes an improved PSO algorithm,called double restriction mutation strategy.When particle tends to leave the search space,in view of the different situation for direction of velocity and position,the strategy controls the particle in the search space effectively,mainly by limiting to updating the position while updating the speed of the particle. This paper lists the performance comparison of four kinds of boundary mutation strategy and this strategy. Experimental studies through five test functions show that the double limit mutation strategy proposed in this paper has faster convergence speed. It is more effective to solve the problem of particle bound. Furthermore,this paper tests the relationship between maximum speed and position on the boundary mutation strategy by experiment. The result shows that the ratio of particles＇ maximum speed and position is inversely proportional to the good or bad degree of the mutation strategy. It provides a basis for the study of boundary mutation strategy.