一种有效的多峰函数优化算法

Effective optimization algorithm for multimodal functions

针对小生境粒子群优化技术中小生境半径等参数选取问题,提出了一种新颖的小生境方法,无须小生境半径等任何参数。通过监视粒子正切函数值的变化,判断各个粒子是否属于同一座山峰,使其追踪所在山峰的最优粒子飞行,进而搜索到每一座山峰极值。算法实现简单,不仅克服了小生境使用中需要参数的弊端,而且解决了粒子群算法只能找到一个解的不足。最后通过对多峰值函数的仿真实验,验证了算法可以准确地找到所有山峰。

This paper proposed a novel niche technique to solve the problem of parameters＇ selection. Any parameters were unnecessary in this method. Through monitoring the changes of tangent value, particles were judged whether there were in the same hill and they would follow the best particle which was in the same hill with them. Each local best solution could be found in this way. The algorithm is easy to realize. So it not only can conquer the shortcoming the limitation of parameters but also only one solution can be found in particle swarm optimization algorithm. The typical numerical simulation results show that the improved algorithm is fairly effective.