Online Demagnetization Fault Recognition for Permanent Magnet Motors Based on the Hall-Effect Analog Sampling

The online demagnetization recognition for permanent magnet motors is of significance but hard for the practical application. To this end, an online fault recognition method is proposed by using analog Hall sensors for the electromagnetic signals. First, a magnetic flux density is reconstructed with the three-dimensional finite element analysis model and sampled signals. Then, the magnetic signal transformation is implemented with the wavelet method, with which low-frequency features are extracted. On this basis, the regional extreme point method is utilized to design the fault classifier and label the fault information including the mounted sides, faulty positions, and severity degrees. Finally, the effectiveness of the proposed online demagnetization recognition method is validated with the simulation and experimental test. The experimental results show that the proposed method can well locate the faulty magnets and identify their fault severity simultaneously; and the overall recognition error is less than 2%. Generally, this proposed online demagnetization fault recognition facilitates the real-time functional security and provides a helpful guidance for the optimal motor control.