基于小波分析和神经网络的电机故障诊断方法研究

在电机故障诊断技术中,电机振动信号最能全面反映电机的运行状态.由于电机振动信号属于非平稳随机信号,传统的傅里叶变换从频域角度进行信号分析,只能说明信号中某频率成分幅值的大小和频率密度,不能检测奇异信号点的时域信息,而且还可能将含有丰富故障信息的微弱信号作为噪声滤去.因此,不能完全满足故障信号特征提取的要求.为解决这一问题,提出一种基于小波分析和神经网络的电机故障诊断方法,该方法采用小波时频分析技术对电机故障振动信号进行消噪滤波,通过小波包分解系数求取频带能量,根据各个频带能量的变化提取故障特征,应用BP神经网络进行故障识别,并采用Matlab仿真软件予以实现.结果表明,该方法不需要建立电机的故障诊断模型,能有效提高电机故障诊断的准确性.

Study of the Fault Diagnosis Method Based on Wavelet Time and Frequency Analysis and the Neural Network in the Motor

In the fault diagnosis technology of motor, the vibration signals can fully reflect the running sta-tus of the motor. As the motor vibration signals are non-stable and random, the signal analysis of the tradi-tional Fourier transform in the frequency domain, can only indicate the amplitude of a certain frequency compo-nent and frequency density, but can't detect the time domain information of the singularity signals and some weak signal with rich fault information is likely to be filtered as noise, so the method above can't fully meet the requirements of fault signals feature extraction. To solve this problem, a motor fault diagnosis method based on wavelet analysis and neural network was presented. This method uses the technology of wavelet time-fre-quency for the noise cancellation and filtering of motor fault diagnosis signals, and strikes the energy of fre-quency band through the coefficient of wavelet packet, gains the fault characteristics from various changes in the energy of each frequency band, and identifies fault through application of BP neural network, and uses Matlab software to realize it. The experimental results show that this method doesn't require establishing the motor fault diagnosis model, and can effectively improve the accuracy of the motor fault diagnosis.