粒子群优化与差分进化混合算法的综述与分类

A SURVEY AND TAXONOMY ON HYBRID ALGORITHMS BASED ON PARTICLE SWARM OPTIMIZATION AND DIFFERENTIAL EVOLUTION

优化算法的性能改进长期以来一直是算法研究者们追求的一个重要目标,对不同算法进行混合以期利用算法的互补优势来获得性能更优异的算法代表了一类典型的设计思想.针对两类基于群体演化的优化算法——粒子群优化(PSO)与差分进化(DE)算法,对基于二者的各种混合算法(DEPSO)进行了系统而全面的综述,并在此基础上提出了一种混合策略分类方法,对现有的各种典型DEPSO算法进行了分类,比较了各种混合策略的异同,并指出了一些新的研究方向和混合设计原则.

Improving the performance of optimization algorithms has long been an important pursuit of researchers.It is a typical design idea and paradigm to combine different optimizers for a synergy of their complementary advantages.Regarding two kinds of population-based evolutionary algorithms,the particle swarm optimizer（PSO） and the differential evolution （DE）,we present a systematic and comprehensive survey on their hybrids（DEPSOs） in the literature and propose a taxonomy of hybridization strategies.Based on the taxonomy, we make a classification of different DEPSOs and analyze their similarities and differences. We also point out some new directions for future research and provide several guidelines for hybridization design of optimizers.