An Improved Model Predictive Torque Control for Switched Reluctance Motor Drive Based on Candidate Switching Vectors Preprocessing

In this paper, an improved model predictive torque control (MPTC) strategy is proposed based on candidate switching vectors (CSVs) preprocessing to minimize the torque ripple and copper loss of switched reluctance motors (SRMs). Firstly, the mechanical angular period is partitioned into six sectors by estimating the demagnetization capability of the outgoing phase online. With this partition, negative torque is avoided and the optimizable region is defined. Secondly, a new commutation rule and switching table are developed to lighten the computational burden at each control period. The CSVs are reduced to 2 or 3, which reduces the computational burden by up to about 78% compared to the conventional MPTC with 9 CSVs. Based on this improvement, the MPTC can be applied to lower-cost platforms. Further, a cost function containing multiple objectives is defined, and the improvement of torque control performance can be achieved easily by traversing the switching table. Compared to direct instantaneous torque control (DITC), this improved MPTC is not limited by hysteresis control and significantly reduces the computational burden compared to conventional MPTC. Finally, relevant simulation and experimental comparison tests are carried out on a three-phase 12/8 SRM. These results demonstrate the effectiveness of the proposed MPTC method and its superiority in improving torque ripple, copper loss and computational burden.