基于BP神经网络算法的异步电机故障诊断系统研究  

Research on Asynchronous Motor Fault Diagnosis System Based on BP Neural Network Algorithm

作　　者：孙吴松[1] SUN Wusong(Department of Mechanical and Electrical Technology,Lu'an Vocational and Technical College,Lu'an 237158,China)

机构地区：[1]六安职业技术学院机电技术系,安徽六安237158

出　　处：《荆楚理工学院学报》2024年第2期1-10,共10页Journal of Jingchu University of Technology

基　　金：安徽省高校学科(专业)拔尖人才学术资助项目(gxbjZD2021111)。

摘　　要：为了确保电机安全可靠地运行,研究了BP神经网络算法对异步电动机进行故障诊断。通过MATLAB平台,分别使用附加动量因子和自适应学习率两种梯度下降法进行网络训练,搭建故障诊断BP网络模型。以MSE值为指标优化最佳隐含层节点数、动量因子与学习率,并通过遗传算法来优化BP网络的初始权值,对故障测试样本进行仿真测试。结果表明,GA-BP网络模型比MF-BP和AG-BP的MSE值更低,仅为0.009163,优化后的诊断预测结果与目标值几乎没有差别。基于遗传算法改进的故障诊断系统模型能够满足异步电动机故障诊断的应用需求。In order to ensure the safe and reliable operation of the motor,the BP neural network algorithm is studied for fault diagnosis of asynchronous motor.Through the MATLAB platform,two gradient descent methods of additional momentum factor and adaptive learning rate are used for network training,and the BP network model for fault diagnosis is built.The MSE value is used as the index to optimize the number of nodes,momentum factor and learning rate of the best hidden layer,and the genetic algorithm is used to optimize the initial weight of the BP network,and the fault test samples are simulated.The results show that the MSE value of GA-BP network model is lower than that of MF-BP and AG-BP,which is only 0.009163.The optimized diagnosis prediction result is almost the same as the target value.The improved fault diagnosis system model based on genetic algorithm can meet the application requirements of asynchronous motor fault diagnosis.

关 键 词：故障诊断 MATLAB BP神经网络 遗传算法 网络优化 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]