融合近邻交互的改进的粒子群优化

Improved Particle Swarm Optimization with Near Neighbor Interaction

提出了一种新的粒子群优化方法――融合近邻交互的粒子群优化算法(Particle Swarm Optimization Combined with Near Neighbor Interaction,NNI_PSO).NNI_PSO在PSO算法的速度更新公式中增加了近邻交互部分,并结合"优胜劣汰",引入动态邻域结构和惯性权值非线性变化.近邻交互有利于粒子快速向全局最优移动,"优胜劣汰"有利于维持种群多样性.将NNI_PSO应用于PSO领域五个著名的基准测试函数,并与其它两个著名的PSO改进算法对比,实验结果证明NNI_PSO在收敛速度和解的精度方面均有明显优势.NNI_PSO不仅提高了PSO算法执行的时间性能,而且有效地缓解了早熟收敛问题.

A new approach of improved particle swarm optimization （PSO）with near neighbor interaction （NNI_PSO） is developed. NNI_PSO incorporates near neighbor interaction into the PSO＇s velocity updating equation, meanwhile, combines with ＂fittest exist＂ by which introduces dynamic neighborhood structure and inertia weight nonlinearly varying into the PSO. The experiments of applying NNI_PSO on five notable benchmark problems, and also the comparing tests between NNi PSO and the other two popular PSO algorithms, demonstrate that the performance of NNI_SPOismuch better tban the other two improved algorithm. Not only can NNI_PSO improve the performance, but also reduce the premature convergence rate effectively in PSO.