Reduced switching frequency operation of power converters using virtual model based MPC

Abstract

Model predictive control has several advantages for power converter applications such as a simple theoretical concept, a straightforward implementation and a high dynamical response regardless the high switching frequency it generates which is one of its main drawbacks. In this paper a virtual first order filter is modeled in the predictive algorithm and its output is included into the cost function in order to modify the switching states selection and reduce the switching frequency. The proposed algorithm generates a reduction of the average switching frequency but it does not deteriorate the dynamical response of predictive control. Additionally, it has a simple implementation because it does not require frequency transformations, the virtual filter can be easily designed and the algorithm does not require additional memory to save the switching pattern to count the commutations. Results that confirm the performance of the proposed algorithm are given.