一种新的多峰值函数优化算法

A Novel Approach to Multimodal Function Optimization

根据对一元多峰值函数的单调区间的分析,提出了一种新的多峰值函数优化算法——形态分析法。该算法根据给定的精度要求,用有穷个离散点确定一元多峰值函数的形态,进而确定其单调递增区间和单调递减区间,一次搜索就可找出函数的所有局部最优解和全局最优解。用不同的多峰值函数进行了仿真实验,并和相关算法进行了比较,结果表明所提出的算法计算量比其它算法小一个数量级,不存在震荡现象。

In this paper,a novel approach to multimodal function optimization Shape Analyzing Algorithm（SAA） is proposed.First,according to the given precision,it determines the shape of one free variable function through finite discrete points,then determines the monotone increasing and monotone decreasing interval of the function.No transcendent knowledge is needed for this algorithm.The function just needs to be searched once and all the local and global optimal solutions can be found,no iteration is needed.The algorithm has been tested to optimize different muhimodal functions,and the simulation results show that the algorithm is valid compared with other algorithms.