Computation-Efficient Model Predictive Control Using Sphere Decoding Algorithm for DTP-PMSMs

Extending the prediction horizon for finite control set model predictive control (FCS-MPC) has shown its advantages on steady-state performance and current distortions. However, its inherently high computational burden brought barriers to implement. Hence, a scheme of sphere decoding algorithm-MPC (SDA-MPC) with preliminary election is proposed for the calculation efficiency and harmonic suppression. The experimental prototype is developed and implemented for dual three phase permanent magnet synchronous motor (DTP-PMSM). Experimental results well agree with the theoretical analysis. In particular, the proposed SDA-MPC scheme significantly reduced the average turnaround time by 21% compared with the exhaustive search algorithm (ESA) under the single step. Moreover, it reduced the ripple in x-y subspace by 14% compared with the conventional SDA with whole control set.