基于小波包神经网络的电机故障诊断分析与研究

由于传统基于傅立叶变换的利用频域对电机故障的信号分析中无法对奇异信号点的时域信息进行检测。针对上述问题,提出基于小波包神经网络的电机故障诊断的方法。结合电机振动的非平稳随机性的特点。利用小波包多分辨率分析方法对电机的采样信号进行分解,提取电机故障状态特征并作为BP神经网络输入样本的特征向量,利用神经网络的自学习和模式识别的特点最终输出电机故障类型。通过MATLAB仿真结果可以证实该方法可行性。

Based on Small Wave Packet neural Network Motor Failure Diagnostic Analysis and Research

The traditional Fourier transform based on use of frequency domain signal analysis of motor faults can not point singular signal detection in time domain information.In response to these problems is proposed based on wavelet packet neural network motor fault diagnosis method.With non-stationary random vibration of motor characteristics.Using multi-resolution analysis of wavelet packet sampling motor signal decomposition,feature extraction and as a motor fault condition BP neural network input feature vector samples,self-learning neural network and pattern recognition features of final output of motor fault type.By MATLAB simulation results confirm the feasibility.