针对多峰函数优化问题先后出现了一系列适应值共享类的遗传算法这些算法。这些都需要事先提供某种信息本文基于事先提供信息的区别提出了一种新的适应值共享类遗传算法的分类方法,并通过一个复杂的标准测试问题对这些算法进行了比较和评价,结果表明在各种算法中,清除算法、动态小生境共享算法和新聚类适应值共享算法具有较高的搜索能力和优化速度。本文的工作对于这些适应值共享类遗传算法的应用和进一步改进具有指导意义。 Aiming at the multi-peak function optimization problem, a series of genetic algorithm of fitness sharing class has appeared. All of these need to provide some kind of information in advance.This paper proposes a new classification method based on the difference of prior information provided by genetic algorithms and compares and evaluates these algorithms through a complex standard test problem. Among the various algorithms, the algorithm of scavenging, the dynamic niche sharing and the algorithm of new clustering fitness sharing have higher search ability and speed of optimization. The work in this paper is instructive for the application and further improvement of these fitness-sharing genetic algorithms.