Gradient-Based Predictive Pulse Pattern Control for Traction Permanent Magnet Synchronous Motor Drives

The Permanent Magnet Synchronous Motor (PMSM) has gained popularity in modern electric traction systems due to its high efficiency, power density, and reliability. Despite these advantages, traction applications demand enhanced control strategies to achieve low torque ripples and fast dynamic response, even during low switching frequency operation. This PhD thesis presents a Gradient-Based Predictive Pulse Pattern Control (GP3C) for PMSMs, combining the low harmonic distortion of Optimized Pulse Patterns (OPPs) with the flexibility and transient performance of Model Predictive Control (MPC). The algorithm optimally manipulates OPPs in real-time, ensuring excellent steady-state performance and fast dynamic response, even in demanding operating conditions. The proposed method is validated through comprehensive testing. Hardware-in-the-loop (HIL) simulations demonstrate the algorithm’s effectiveness across diverse scenarios, including flux-weakening, dc-link disturbances, and fast acceleration, outperforming traditional Field-Oriented Control (FOC). Experimental results on a laboratory PMSM test bench further confirm the superior harmonic and transient performance of the developed algorithm.