Research On Fault Diagnosis Method Of Brushless Dc Motor Driver

Brushless DC motor (BLDCM) is playing more important effect in industrial, commercial, aviation, military and other fields. Through diagnosing and analyzing the common fault, we can discover earlier and prevent in time and decrease the damage in a shut. The position which BLDCM is often out of order often occurs in the drive, that is inverter. The power transistor and control loop of the inverter always go wrong, and the fault of power transistor account for 82.5 percent of the whole drive system. So the paper mainly research the fault diagnosis and locating of the power transistor of inverter. The main content of paper is following. Wavelet analysis is a time-frequency analysis method following the FourierTransform and Short-Time Fourier Transform, this method is suitable for the fault diagnosis. Through contrast the wavelet analysis with Fourier Transform to get the data of fault feature, we account the advantage of wavelet analysis. Because of different wavelet function solve the same problem that get different result. In study I use the method comparison to obtain suitable wavelet function for getting data of fault feature. RBF neural network have merit of good local quality and simple structure etc. RBF neural network solve first problem is choosing the parameter of hidden layer. In study, contrasting clustering approach with ordinary least squares (OLS), obtain the OLS have character of running quickly, smaller network size and classifying better. So using the OLS train network in paper. Combining wavelet analysis with RBF neural network apply to the fault diagnosis of inverter have better effect.Because of the fault diagnosis method apply to the open-loop controlled inverter, in the paper, using the fault diagnosis of combination of wavelet multi-resolution analysis and RBF neural network apply to the brushless DC motor in order to practical value.