Diagnosis of Partial Demagnetization in Permanent Magnet Synchronous Machine using Wavelet Packet Transform

Demagnetization is a common fault that threatens the permanent magnet machine’s health. The main purpose of this paper is to estimate demagnetization severity in the PM machines to monitor the level of magnet health and prevent the spread of fault to other components of the machine. In this paper, a time-frequency detection algorithm has been designed based on a signal-based method called wavelet packet transform. The purpose of this algorithm is to estimate a straight line using least square in a new plane for evaluating the fault severity. The ability of this method has been tested on a 3 kW simulated synchronous linear generator in finite element method software and an experimental 1 kW radial flux permanent magnet synchronous machine. In addition, by applying this method to the experimental machine, it was determined that this method has a proper ability in different types of permanent magnet synchronous machines with varying speeds.