基于多准则寻优策略的差分进化算法

Differential Evolution Algorithm Based on Multi-criteria Optimization Strategy

在差分进化算法的基础上,提出一种基于多准则寻优策略的改进差分进化算法。该算法可以动态调整变异因子和交叉概率,基于文中提出的多准则寻优策略,通过个体适应度、个体间距离等评价指标判断个体的优劣程度,并且可以降低种群的高密度程度,增强种群多样性。这种判断机制可以有效避免种群过早收敛,易陷入局部最优的风险。通过具体的测试函数对算法进行测试,并与标准差分进化算法进行比较,结果显示算法寻优效果较好,可以较快地得到全局最优解。

Based on differential evolution algorithm,this paper proposed an improved differential evolution algorithm with multi-criteria optimization strategy.This algorithm can adjust the mutation and crossover parameters dynamically,and evaluate individuals＇ grade of excellence with the evaluation indexes like individual fitness and distance between individuals on the basis of the proposed multi-criteria optimization strategies.It can also reduce the degree of population density and enhance population diversity.Although this evaluation mechanism can effectively avoid premature convergence of population,it tends to result in local optimum risk.After it was tested by some test functionsand compared with standard differential evolution algorithm,this improved algorithm was found to have the best optimization effect,being able to obtain the global optimal solution quickly.