基于极端学习机的开关磁阻电机故障诊断研究

Switched Reluctance Motor Fault Diagnosis Based on Extreme Learning Machine

该文主要提出并讨论了一种基于稀疏编码的极端学习机算法的开关磁阻电机的故障诊断的方法。通过神经网络的初始化过程成功地将极端学习机算法与隐层神经元的稀疏编码结合在一起。实验结果表明该文提出的算法可以对电机一相绕组开、短路故障以及正常运转的状态进行准确、快捷的识别,有效地避免神经网络结构过于复杂所导致的过拟合现象。

In this paper,a new method of switched reluctance motor fault diagnosis based on sparse coded extreme learning machine is proposed and discussed.In this method,extreme learning machine is combined with sparse coding of the hidden layer neurons through an initialization process of neural network.The experimental results demonstrate that the new proposed method can not only deal with the fault diagnosis problem accurately and rapidly,but also be enable to avoid the over-fitting phenomenon caused by complexity of neural network structure.