基于记忆策略的动态离子运动优化算法

Dynamic ion motion optimization algorithm based on memory strategy

针对动态优化问题求解时普遍存在的多样性保持不够、寻优能力不足等问题,提出了一种基于记忆策略的动态离子运动优化算法。该算法引入了全局最优个体和斥力的作用,改进了固体阶段的启动条件和更新方式;提出了根据概率选取存储记忆个体,构建正、反向记忆种群;依据余弦相似度排序规则,利用正、反向记忆种群更新进化种群,加快进化种群跟踪最优解的能力;并提出基于余弦相似度排序的环境变化后增加种群多样性的方法。将本文算法与近年效果较好的4种动态优化算法在国际上通用的动态测试函数DF1和MPB移动峰问题上进行测试,仿真结果表明,本文算法在求解精度、收敛速度及稳定性上都优于其他对比算法。

To solve the problems of insufficient diversity and searching ability in solving dynamic optimization problems,a dynamic ion motion optimization algorithm based on memory strategy(MDIMO)is proposed.The algorithm introduces the role of global optimal individuals and repulsion.The start-up conditions and renewal methods of solid phase are improved.The selection of memory individuals based on probability is proposed,and positive and reverse memory populations(PRMPs)re constructed.According to the cosine similarity ranking rule,The evolutionary populations are renewed by using PRMPs,and the ability of evolutionary populations to track optimal solutions is accelerated.A method of increasing diversity after environmental change based on cosine similarity ranking is proposed.The proposed algorithm is tested on the international dynamic test function DF1 and MPB mobile peak problem,and compared with foure dynamic optimization algorithms which have better effect in rcent years.Simulation results show that the proposed algorithm is superior to the other algorithms in solving accuracy,convergence speed and stability.