采用小波变换与RBF神经网络优化的电机故障模糊诊断系统

针对特征信号淹没于噪声信号的情况,采用Morlet小波分析实现了对原始电流特征信号的降噪.同时,采用基于RBF神经网络的最优化插值与具有频谱细化特性的CZT分析,提升了频谱分辨率,充分展现了发生故障时电流特征信号的频谱细节,为电机故障诊断系统提供了可靠的诊断依据.建立了基于改进型BP神经网络的电机故障模糊诊断系统,抽象出了偏心故障的诊断规则.实测结果表明,该系统能够可靠地诊断电机的偏心故障.

Fuzzy Diagnostic System for Induction Motor Based on Wavelet Analysis and RBF Neural Network

In view of the situation that characteristic signals may be submerged in noise signals,wavelet analysis was adopted to reduce the amount of noise signal in the original characteristic signals of the current.Meanwhile,by using the optimization interpolation based on RBF Neural Network and CTZ Analysis which possesses refined spectrum characteristics,the spectral resolution was improved,fully reflecting the spectrum details of current’s characteristic signal in the situation of breakdown,thus,providing reliable basis for the motor diagnostic system.A fuzzy diagnostic system for induction motor was established based on improved BP Neural Network.Besides the regulation for the diagnosis of eccentric fault was summarized.Experimental results indicate that this system can reliably diagnose the eccentric fault of induction motor.