Motor Bearing Fault Prediction Using Motor Current Signature Analysis

Bearing fault prognosis and remaining useful life prediction is a key problem in condition based maintenance.The motor stator current contains abundant information which reflects the working condition of motors.In the light of the remaining useful life prediction of auxiliary motor bearings of the HXN3 type locomotive,an efficient fault forecasting procedure using motor current signature and combined prediction model is presented.In this procedure,the condition monitoring is performed based on the on-line motor current measurements,and further,the fault quantification is performed by the grey renewal-particle filter forecasting model.The contributions are made as follows:Stator current acts as an excellent transducer for detecting faults in motors.The bearing faults induce the relating frequency components in stator current,and then the stator current spectrum changes slightly.According to this this principle,the quantitative indices which are sensitive to the current spectrum can be used to describe the development of faults.Then,three quantitative indices,i.e.the spectral entropy,the wavelet packet energy entropy and the frequency standard deviation were extracted from the motor current signal.Experimental results show that the three parameters increase monotonically with motor bearing faults.For precise and reliable fault prediction,a combined prediction model,i.e.the grey renewal-particle filter forecasting model is proposed.In this procedure,a dynamic spatial state model is established first using the grey renewal model.Then,the model is brought into the particle filter algorithm to establish a combined prediction model.Since this forecasting procedure take advantages of two different models,the changes in each index can be described accurately.The stator current signals of the auxiliary motors of HXN3 type locomotives were sampled.The diagnostic indicators,such as the wavelet packet energy entropy,the spectral entropy and the frequency standard deviation were extracted and the proposed combined prediction models were established.The experimental results show that the proposed forecasting procedure considers several diagnostic,therefore it is highly suitable to quantify the faults development and to determine the bearing remaining useful life.