基于遗传算法收敛特性的停机准则

关于遗传算法及改进遗传算法收敛问题的研究表明,理论上分析各类不同结构遗传算法的收敛性是可行的.对已被证明收敛的遗传算法,在计算过程中最突出的问题便是如何判断当前计算结果已经到达最优解从而停止迭代.文中从遗传算法收敛特点及不同种群中最优个体适应值的一致性、种群的多样性出发,提出判断算法自动停止迭代的依据.

Stop Criterion Based on the Convergence Properties of GA

The study on the convergence problems of genetic algorithms （GA） and improved genetic al- gorithms （IGA） at present shows that it＇s feasible to analyze the convergence of GA with different structural models theoretically. To the GA which convergence properties had been established, the most prominent problem in calculation process was how to estimate the current calculated results had been reaching the optimal value and then stopping the iterative process of the algorithms. An auto- stop criterion based on the convergence properties of GA, the consistency of optimum individual fit- ness in different populations and the diversity of population was proposed in this paper.