具有最优控制参数的差分进化算法

Differential Evolution Algorithm with Optimal Control Parameters

针对差分进化算法中控制参数不易选择的问题,提出一种具有最优控制参数的差分进化算法.将控制参数选择问题转化为函数优化问题,采用主从结构的算法模式,在主程序中对控制参数进行寻优,在子程序中对目标函数进行优化,实现了控制参数和目标函数的同步优化.基准函数测试表明,该算法具有更快的收敛速度和更高的搜索精度.该方法对于差分进化算法参数选择具有指导意义.

To overcome the shortcoming that control parameters were hard to be selected in differential evolution(DE)algorithm,a novel differential evolution(OCPDE)algorithm with optimal control parameters was proposed.The problem of selecting control parameters was transformed into the problem of optimizing function.The main-subordinate structure scheme was used in OCPDE.Control parameters were optimized by DE in main program and objective function was also optimized by DE in subordinate program,which implemented optimization of control parameters and objective function simultaneously.Benchmark functions were selected as testing functions,and experimental results show the OCPDE is faster convergence speed and higher search precision in comparison with other DE.This method has guiding significance for parameter selection of differential evolution algorithm.