遗传算法优化速度的改进

分析了传统变异算子的不足,提出用二元变异算子代替传统的变异算子,并讨论了它在克服早熟收敛方面的作用.同时,针对二进制编码的遗传算法的特点,提出了解码算法的隐式实现方案,使得遗传算法的寻优时间缩短6～50倍.实验从多方面对二元变异算子的遗传算法进行性能测试,结果表明，改进型算法收敛快,参数鲁棒性好,能有效地克服“早熟”收敛.通过改进变异算子和解码算法,遗传算法的优化速度得到了很大的提高.

Improving Optimization Speed for Genetic Algorithms

The disadvantage of the traditional mutation operator of GAs was analyzed in this paper, and a DMO (dyadic mutation operator) was presented to take the place of the traditional one. The function of DMO to prevent premature convergence was also discussed. Meanwhile, according to the features of binary-based GAs, an implicit implementation for decoding the chromosomes for GAs was presented so that the run time of the improved program for GAs was shortened by 6～50 times compared with the original one. The performance of the genetic algorithm is tested based on the DMO (GADMO) in several aspects. The experimental results show that the GADMO can converge quickly and its robustness of parameters is strong. The GADMO can prevent the premature convergence effectively. By improving the mutation operator and the decoding algorithm, the optimization speed of GA is speeded up greatly.