遗传算法中模式的竞争度研究

Research on Mode Competitive Degree in Genetic Algorithm

针对Goldberg提出的利用竞争模式对欺骗问题进行度量所存在的弊端,提出模式竞争度的度量新方法,定义竞争函数及竞争度,给出模式竞争度的性质,证明模式竞争度和重要模式之间的关系,利用模式竞争度对2个典型的模式欺骗性与GA欺骗性不符的函数的遗传算法运行机理做了合理解释,表明模式竞争度作为新度量方法的合理性。

Aiming at for the disadvantages of using competitive mode to analyze deceptive functions which is promoted by Goldberg, this paper gives a new measuring method from a new angle. It defines competitive function and competitive degree, gives some characteristics of it, proves the connections between competitive degree and important mode, and by using competitive degree rationally explains how the Genetic Algorithm（GA） solves two representative functions which mode deception is not in correspond with GA＇s deception. It proves that competitive degree as a new measuring method is rational.