Improved Bearing Fault Modeling and Current Signature Analysis for Permanent Magnet Synchronous Motors

Bearing fault detection and severity classification is crucial for improving the reliability of permanent magnet synchronous machines (PMSMs). These failures are critical in significant applications, which include electrified transportation and aerospace applications. The current signal-based diagnostics methods are increasingly popular over traditional vibration-based methods. This is because of its comparative advantages, such as lower cost and ease of implementation. The bearing condition monitoring techniques rely exclusively on experimental validation, which can be costly and time-consuming. The lack of standard datasets with current data also hinders further research. This paper presents a generalized model-based approach to overcome these challenges. It combines mathematical modeling and simulation to study and emulate torque disturbances in PMSMs resulting from localized faults. The phase modulation in the stator current under these faults is induced by disturbances in load torque. This is an important aspect of condition monitoring. The proposed approach is validated through simulations using the MATLAB/Simulink model and experimental analyses, which assess major localized faults under dynamic operating conditions and measurement noise utilizing frequency domain analysis. It demonstrates the potential use of this approach to develop reliable condition-monitoring techniques, integrating simulation data with limited experimental data to create robust datasets for data-intensive machine learning-based fault diagnosis while addressing the challenges in laboratory testing.