摘要
首先分析了遗传算子对遗传算法的统计学特性的影响,并对十进制遗传算法进行了一定的改进。在此基础上,提出了算法中参数的选择依据,然后将其与神经网络的BP算法相结合,对神经网络的权值进行全局搜索。仿真实验结果表明。
The influence of genetic operators on the statistic characteristics of genetic algorithm is studied,then the real GAs has been modified.Based on this,the select of parameters in the GAs algorithm is introduced.Then combined with the back-propgation algorithms of neural networks,this algorithm is used to search the best weights of neural networks in whole searching space.The results of the simulation experiment shows that the algarithm has good effect.
出处
《电气传动自动化》
2000年第6期35-38,共4页
Electric Drive Automation
关键词
遗传算法
神经网络
模式
交叉变异
genetic algorithms
neural networks
mode
crossover mutation
