一种新的求解多峰函数优化问题的动态演化算法

A Novel Dynamic-Population Based Evolutionary Algorithm for Multimodal Function Optimization

本文给出了一种新的求解多峰函数优化问题的定义:定位所有的极值点,包括全局的峰值点和局部的峰值点。传统的演化算法框架都是群体固定的演化迭代过程,对求解多峰函数优化问题时由于无法事先得知峰值点的个数而很难确定合适的群体大小,影响了算法的效率。提出一种群体动态可调的演化方式,使得初始群体大小可任意指定,在演化过程中通过聚集和按比例引入新个体两个过程而动态变化。实验表明,该算法能尽可能多地定位峰值点。

The traditional evolutionary algorithm with a fixed-size population is not suitable especially for solving multimodal function optimization because it's impossible to know the number of solution in advance and hence it's difficult to specify a suitable size of population. In this paper, a novel algorithm with dynamic population is presented. In the process of evolution, the size of population is tuned by a aggregation and introduction of new individuals. A initial experiment is given.