一种基于JADE改进的差分演化算法

差分演化算法有局部搜索能力不足、容易跌入局部最优等缺点,其搜索性能主要依赖于对杂交概率和缩放因子的设置。为了改善上述缺陷,对带归档的自适应差分演化算法JADE进行深入的研究与分析,提出了改进的自适应差分演化算法ZJADE。该算法采用斜帐篷混沌映射函数初始化种群,在每次迭代中为每个个体分别产生满足正态分布、柯西分布的杂交概率和满足正态分布的缩放因子,并且记录成功变异个体的杂交概率和缩放因子,引入统计杂交概率,采用两种策略自适应地更新杂交概率。在13个经典测试函数上将ZJADE算法与多种经典自适应差分演化算法进行对比,实验结果表明,ZJADE算法在解的精度与收敛速度上更优,具有更好的搜索性能。

An improved differential evolution algorithm based on JADE

Differential evolution algorithms are weak in local searching and easy to dropping into the local optimal solutions at the same time. The search performance of these algorithms is mainly based on the parameter setting of their crossover probability and mutation factors. To improve the above shortcomings of differential evolution algorithms, we propose an adaptive differential evolution algorithm called ZJADE on the basis of in depth research and analysis of the adaptive differential evolution with optional external archive (JADE). Skew tent chaotic mapping is used to initialize the population in order to generate uniformly dispersed population. During each generation, the crossover probability of each individual is generated according to the normal distribution and the Cauchy distribution while the mutation factors are independently generated according to the normal distribution. The crossover probability and mutation factors of successful individuals are saved, and the statistical crossover probability is employed. The ZJADE algorithm is compared with multiple state of the art adaptive differential evolution algorithms through thirteen classical test functions. The results show that the ZJADE obtains better solution accuracy and quicker convergence speed, thus having a better search performance.