Reduction of Computational Complexity in Modulated Model-Predictive Control for Synchronous Reluctance Motor Drives

Abstract

This paper introduces a novel geometric approach to significantly reduce the computational burden of modulated predictive controllers while maintaining the same steady-state performance and satisfactory dynamic behavior. The proposed geometric method leverages the symmetric properties of the active vectors with respect to the zero vectors in two-level inverters. In addition, the structure of the controller is designed to include the integral of error terms, ensuring zero steadystate tracking error. Several operating points are considered and compared with respect to standard modulated model-predictive control approaches, showing similar steady-state performance with a reduced computational effort (about 50% of the classical implementation execution time). This enables the use of more complex power electronics conversion system, which requires higher number of predictions, or to increase the switching frequency in traditional two level inverters, without compromising the steady state performance of the proposed controller. The effectiveness of the proposed approach is demonstrated with a synchronous reluctance motor drive application.