Predictive Control Scheme With Adaptive Overmodulation for a Five-Leg VSI Driving Dual PMSMs

A dual permanent magnet synchronous motor (PMSM) system, driven by a five-leg voltage-source inverter, is considered a feasible option for low-cost and fault-tolerant applications. The finite-control-set model predictive control schemes without modulators generate a heavy computational burden and larger current ripple. In this article, an improved predictive control method with a space vector modulation (SVM) scheme is proposed for independent control of a dual-motor system. First, the expected voltage vectors (VVs) of dual motors are obtained with a deadbeat solution, which yields a faster dynamic response. Then, in order to reduce the current ripple and eliminate the candidate VV traversal process, an improved modulator with adaptive overmodulation is adopted to synthesize the expected VVs. The presented SVM method solves the problem caused by the conflicting switching-state demands of the common leg by dual PMSMs. Moreover, an extensive comparison is performed among the proposed control method, the existing model-based control methods, and the field-oriented control method with a proportional–integral controller. All methods are experimentally carried out in steady and transient states. The test results indicate that the proposed control scheme with adaptive overmodulation can effectively reduce current ripple, improve voltage utilization, and preserve good dynamic performance.