Computational-Efficient Model Predictive Torque Control for Switched Reluctance Machines With Linear-Model-Based Equivalent Transformations

In this article, a novel model predictive torque control (MPTC) method for switched reluctance machines (SRMs) is proposed based on the equivalent linear SRM model and the improved switching table. First, an improved switching table with only six switching states is developed based on the inductance characteristics. The adoption of this improved switching table not only reduces the computational burden by 25% but also improves the torque control performance and system efficiency at high-speed region. However, the look-up-table (LUT) based MPTC methods suffer from occupying numerous storage space. To ease this issue, the simple linear SRM model is utilized. The flux-linkage and torque equivalent transformations are introduced to address the difference between the linear and nonlinear SRM models. With these transformations, the MPTC realized on the linear SRM model is proposed. Compared to 1530 storage units consumption to store three two-dimensional (2-D) LUTs in the LUT-based MPTC methods, the proposed method only occupies 228 storage units. Experimental results on an 8/6 SRM setup verify that the proposed method effectively eliminates the commutation torque ripple with lower execution time, less storage space, and improves the torque control performance at a high-speed range compared with the conventional LUT-based method.