摘要
为了实现无刷直流电机控制器中IGBT开路故障的定位,该文采用短时傅里叶分析结合卷积神经网络的故障诊断方法。首先基于无刷直流电机的数学模型仿真双闭环控制系统,采样IGBT不同故障状态下电机输出的三相电流。其次通过短时傅里叶变换进行时频变换,获得电流故障信号的时频图像。最后利用卷积神经网络对时频特征图像进行故障诊断和定位,结果表明该方法能够准确识别IGBT的开路故障。
In order to realize the location of IGBT open circuit fault in Brushless DC motor controller,this paper adopts the fault di⁃agnosis method of short-time Fourier analysis combined with convolution neural network.Firstly,based on the mathematical model of Brushless DC motor,the double closed-loop control system is simulated,and the three-phase current of motor output under dif⁃ferent fault states of IGBT is sampled.Secondly,the time-frequency image of current fault signal is obtained by short-time Fourier transform.Finally,convolution neural network is used to diagnose and locate the fault of time-frequency feature image.The results show that the method can accurately identify the open circuit fault of IGBT.
作者
邹爽
朱建光
ZOU Shuang;ZHU Jian-guang(Shenyang University of Technology,Shenyang 110870,China)
出处
《电脑知识与技术》
2021年第30期148-150,共3页
Computer Knowledge and Technology