双吸引子多群体粒子群算法解决动态优化问题(英文)

Bi-attractor Multi-population Particle Swarm Algorithm for Dynamic Optimization Problems

提出了一种新的"双吸引子多群体粒子群优化算法"(BMPSO)。与传统的粒子群优化算法(PSO)相比,BMPSO的主要特点是它使用了两个"群体吸引子"和两种搜索粒子。两种搜索粒子具有不同范围的搜索特性,一种利于进行全局搜索而另一种利于进行局部搜索。并且通过引入一种新的"传递"机制,两部分粒子可以更有效地共享搜索信息。实验表明,BMPSO算法在Moving Peaks Benchmark(MPB)测试问题上具有很好的性能表现。

A new Bi-attractor Multi-population Particle Swarm algorithm （BMPSO） was proposed.Compared with traditional particle swarm algorithms （PSO）,BMPSO uses two types of search particles and two swarm attractors.The two types of search particles focus on different search scopes：one focuses on local search and the other focuses on global search.Furthermore,by adopting a new transfer mechanism,the two types of particles can share their experienced search information more effectively.Experimental analysis shows that BMPSO performs pretty well on the Moving Peaks Benchmark problem （MPB）.