遗传算法的随机摄动法

Random Perturbation Method of Genetic Algorithms

为避免遗传算法在进化计算过程中失去种群多样性并陷入局部最优解 ,提高其全局优化搜索的运行效率和精度 ,提出一种遗传算法的改进方法——随机摄动算法 .它是将复制、交叉、变异后群体中的各个体在确定的某个邻域随机摄动 ,如引起适应度值增加 ,则接受随机摄动操作 ,否则以变异概率接受随机摄动操作 .然后以能量函数的概念对其运行机理进行了说明 .算例表明 ,本文所提出的随机摄动算法易于实施 ,且十分有效 .

An improvement method of genetic algorithms——a random perturbation method is proposed in this paper, to avoid losing population variety which leads to fall into a local optimal solution, and to improve its running efficiency and precision of overall optimization searching during evolution calculation. Namely, the individuals of the population after reproduction, crossover and mutation operation are randomly perturbed within a certain neighborhood. The perturbation operation is accepted if it causes the fitness value to go up ,otherwise,accepted at the mutation probability. Besides, a brief expounding of its running mechanism is given by the concept of energy function. The examples show that the random perturbation method is both easy to be carried out and extremely efficient.