一种新的双种群PSO-DE混合算法

摘要

给出一种新的粒子群算法和差分进化算法相结合的混合算法.该算法基于一种双种群进化策略,其中一个种群由粒子群算法进化,另一种群由差分进化算法进化.此外,采用一种信息分享机制,在算法的进化过程中2个种群中的个体可以实现协同进化.为了进一步提高混合算法的性能,在差分进化算法中融入一种线性递减加权策略的变异操作和指数递增交叉概率算子.通过4个标准测试函数的测试结果表明文中提出的混合算法是一种收敛速度快、求解精度高、鲁棒性较强的全局优化算法.%In this paper,a new hybrid algorithm of PSO and DE algorithm is given. The algorithm base on a dual populations evolutionary strategy, a population is evolved by the PSO,and the other is e-volved by DE Algorithm. In addition, we introduce a mechanism to share information, the individual of two populations can achieved co-evolution in the evolutionary process. In order to improve global optimization ability of the hybrid algorithm, a new type of mutation of a linear decline weighted strategy and exponent increased crossover probability operator is incorporated in differential evolution algorithm. Four benchmark functions' results show that new algorithm is a kind global optimization algorithm of fast convergence, high accuracy and more robust.