Model Predictive Control for Electric Drive Systems

Abstract

Model Predictive Control (MPC) has emerged as an influential control strategy for electric drive systems, offering enhanced dynamic performance, constraint handling, and multi-objective optimization capabilities. Unlike traditional techniques such as Field-Oriented Control (FOC) or Direct Torque Control (DTC), MPC optimizes control actions by predicting future behavior over a horizon and minimizing a cost function. Pre-2018 implementations—including Finite-Control-Set MPC (FCS-MPC) and Continuous-Time MPC (CT-MPC)—address torque, flux, and current regulation directly via voltage vector selection, offering rapid responses and reduced ripple. Key advancements include torque control with minimized ripple in Induction Machines ([IET, 2015]), low switching frequency MPCC for Permanent Magnet Synchronous Motors (PM-SMs), and sensorless predictive torque control of Induction Machines (IMs) encompassing reactive power regulation ([2018 thesis]). Challenges remain in computational demand, cost weighting factor tuning, and system model fidelity. Comparative studies underline MPC’s ability to manage converter constraints and nonlinearities more intuitively compared to PWM-based FOC. This paper reviews foundational MPC formulations in drive applications up to 2017, analyzes experimental achievements, and synthesizes trade-offs in performance, implementation complexity, and robustness. In conclusion, we outline future directions such as reduced-complexity algorithms, explicit MPC, and multi-phase drive extensions.