A Generalized Current Residual-Based Diagnosis and Classification Method for Interturn Fault and High-Resistance Connection Fault in Model Predictive Controlled PMSMs

Model predictive control (MPC) scheme is widely applied in motor drives due to its fast response and flexibility. However, motor fault diagnosis methods under MPC remain under-explored. The residual current has been explored for interturn fault (ITF) diagnosis. Considering current residuals, the similarity in fault features between ITF and another fault, i.e., high-resistance connection (HRC), has not been fully considered. The classification of these two types of faults has not been sufficiently studied either. This article proposes a generalized current residual-based method for the diagnosis and classification of ITF and HRC in PMSMs under MPC. By leveraging the rolling optimization feature of MPC, the proposed method first obtains the current residuals in both continuous and discrete time domains by comparing the measured currents with the MPC-predicted currents in the $\alpha\beta$-axis. Next, fault diagnosis, location, and classification indicators are constructed using the difference and ratio of the current residuals, which can enable fault diagnosis, location, and classification within a unified framework. This method can be fully implemented online without the additional observers, hardware modifications, and alterations to the control structure. The effectiveness of the proposed fault diagnosis and classification method is verified by extensive experiments.