杂交算子的配对与协同性

Pairing of crossovers and its synergy

在遗传算法中两个或多个杂交算子的适当组合能够产生协同效应,从而可以提高算法的搜索性能。为分析其机理并提出协同组合方法,对两个杂交算子的协同组合问题进行了研究。结果表明,组合中算子对群体多样性的调节和利用是产生协同效应的关键。进而提出了两个杂交算子协同组合的方法,即组合中要有能使群体收敛快的和收敛慢的算子,以调节群体多样性;要有求解质量较好的算子,以利用群体多样性求出更好的解;通过调整两个算子的比例可调节群体收敛速度以及平衡好算子的求解质量和收敛速度之间的关系,使得算子组合形成协同求解能力。为获得更好的协同效果,还要考虑算子组合的问题相关性。详尽的数值实验结果表明了分析的正确性和所提出的协同组合方法的有效性。

In genetic algorithms careful combination of two or more different crossovers can produce synergy, thus improving genetic algorithm performance. To analyse the mechanism and provide a method of synergic combination of crossovers, we examine the synergie combination of two cross overs. It is shown that the key for the combination to produce synergy lies in the adjustment and us age of the population diversity of the crossovers. Thereby the method of synergically combining two crossovers is proposed. The combination should be composed of two crossovers, one of which makes the population converging quickly while the other slowly, so as to control the diversity of the popula tion. Of cause it should contain a crossover with good solution quality which makes use of the popu lation diversity to produce a better solution. Meanwhile, by adjusting the proportion of the two cross overs, the convergence rate of the population can be controlled while the relationship can be well balanced between the solution quality of the crossovers and the convergence rate of the population, enabling the combination to form a synergic solution capability. In addition, to obtain better synergy results the correlation between the crossover combination and the problem to be solved should be considered. Detailed numeric experimental results confirm the correctness of the analysis and the ef fectiveness of the proposed method of synergic combination.