Efficient Maximum Torque Per Ampere (MTPA) Control of Interior PMSM Using Sparse Bayesian Based Offline Data-Driven Model With Online Magnet Temperature Compensation

The maximum torque per ampere (MTPA) is popular control strategy for interior permanent synchronous machines (PMSMs) and MTPA point is dependent on the magnetic saturation and magnet temperature. This article proposes a novel MTPA control method combining offline model and online compensation model for interior PMSM control. In the proposed approach, an offline sparse Bayesian based data driven model is derived from the machine equations to consider magnetic saturation, and an online compensation model is proposed to compensate the magnet temperature. The MTPA point can be derived by combining both the offline and online models, in which both saturation and temperature effects are considered to ensure the performance of MTPA point tracking. Compared with the offline methods, the proposed approach employs the sparse vector to represent the MTPA model with less computation and memory consumption and considers the temperature effect with better robustness. Compared with the online methods, the proposed approach only compensates the offline model with online temperature effect, which is less sensitive to noise and uncertainties and involves less computation. The proposed approach is validated with comparisons and experiments on a laboratory interior PMSM drives.