On-Line Detection and Classification of Permanent Magnet Demagnetization using Observed Flux Linkage Signals

Demagnetization in permanent magnet synchronous motor (PMSM), due to factors like high temperature or reverse magnetic field can lead to increased loss, torque fluctuations, and potential instability of the system. Detecting this demagnetization through online methods allows for the identification of permanent magnet (PM) weakening at the initial stage. In this paper, an online demagnetization detection technique is proposed by using a flux observer. The flux observer is constructed based on the voltage equations of PMSM in dq reference frame. It is used to estimate the fundamental and harmonic components of the flux linkages corresponding to uniform and partial demagnetizations. Fault indicators are established based on the estimated amplitude of special harmonics. Therefore, the proposed approach not only identifies PM demagnetization fault and its severity, but also distinguishes between different fault types. Additionally, the proposed method has less computational complexity and few variables, it is easy to implement without any extra hardware. The proposed approach is validated through simulations with the observer applied to estimate the flux linkages from a precise PMSM finite element model under no-load and loaded conditions with different demagnetization scenarios.