Parallel block coordinate descent algorithm for model predictive control in power electronics

Solving long prediction horizon model predictive control (MPC) problems in real time is a big challenge in power electronics applications. To address this challenge, this work presents the parallel block coordinate descent (BCD) algorithm as an efficient and simple way to solve MPC problems with a structure that is common across many power electronics applications. To showcase the effectiveness and simplicity of the parallel BCD, it is applied to the output voltage control of a buck converter and to the current control of an RL load, showing the low computational complexity and the simple concept behind each iteration of the algorithm.