A simple and effective strategy to reduce switching losses under FS-MPC based on dynamically changing Voronoi diagrams

Finite Set Model Predictive Control (FS-MPC) is a powerful and popular strategy to control power electronic converters. One important aspect of FS-MPC is to limit switching losses while tracking a given reference. This is typically achieved by including in the cost function an explicit measure of switching losses alongside the tracking error. Appropriate weighting coefficients to these two different measures are typically assigned by trial and error. The paper proposes a novel and very effective way to reduce switching losses without penalising them explicitly in the cost function. This is achieved by virtue of fictitious voltage states, which change their positions depending on the currently chosen switching state. An implementation algorithm proposed in the paper is based on principles of computational geometry, specifically, Voronoi diagrams. The algorithm is very efficient and poses almost no additional computational burden compared to standard quadratic cost function. Its application significantly reduces switching losses of FS-MPC without a major penalty in terms of its harmonic distortion. Moreover, the paper shows that the proposed algorithm can be easily combined with other advances recently developed around MPC. The paper presents extensive simulation results showing the advantages of the proposed method and comparing the performance of its various options.