摘要
针对BP神经网络收敛速度慢、易陷入局部极小的缺点,将具有全局搜索能力的遗传算法引入到神经网络的权值优化中。遗传算法优化神经网络模型时,参数选取直接关系到模型优化的效率,在给出一种遗传算法的基础上对相关参数进行了研究分析。并采用Matlab软件编程实现算法,把该算法应用到XOR问题求解中,显示出GA-BP算法的优越性,并通过磨机故障诊断实例验证了算法的有效性。
To overcome shortcomings of the BP neural network, this paper uses the global random hunting function of the genetic algorithm to train coefficients of neural network. In the optimization process of neural network model based on genetic algorithm, the optimization efficiency of the model is closely related with the choice of parameters. By giving a kind of genetic algorithm, the relevant parameters are studied and analyzed. And this network is used to complete XOR calculation. Results show that the algorithm has great advantages and experiments on fault diagnosis show that it has practical use to a certain extent.
出处
《北京信息科技大学学报(自然科学版)》
2009年第3期35-38,共4页
Journal of Beijing Information Science and Technology University
基金
北京市自然科学基金项目(3042006)
北京市教育委员会科技项目(82052002)