Model Predictive Torque Control of Permanent Magnet Fault-Tolerant Rim-Driven Motor Based on Reverse Prediction

The integration of model predictive control (MPC) and direct torque control (DTC) has become an attractive solution for high performance torque regulation in fault-tolerant permanent magnet rim driven motor (FTPM-RDM). The motor features a unique independent H-bridge inverter topology with spatially decoupled windings, which provides inherent fault tolerance but also poses challenges due to its large voltage vector space. Traditional model predictive torque control (MPTC) with voltage vector preselection requires repeated screening of all candidates, resulting in high computational complexity and limited adaptability under faulted conditions. This paper proposes a reverse-prediction-based MPTC strategy that derives the required voltage vector amplitude and angle in real time and dynamically reconstructs the feasible set. By doing so, it avoids static lookup tables and reduces the number of candidate vectors from 61 to typically 2-3, achieving a 57% reduction in execution time. Under single-phase open-circuit faults, the strategy reconstructs the voltage vector space using only the healthy phases, thereby suppressing torque ripple and ensuring rapid recovery. Experimental validation verifies the superior performance of the proposed method: it reduces computation time to 31 μs, ensuring real-time operation within the 100 μs control period, and under fault conditions, it achieves recovery within 0.26 s while cutting torque ripple to 11.27%, compared with 25.96% without the proposed strategy. These results confirm that the proposed control strategy combines improved computational efficiency with robust fault-tolerant performance, offering clear advantages over traditional MPTC.