An SOR-like method for fast model predictive control

Abstract

This project proposes an iterative first-order gradient method for solving convex quadratic programming problems, which involve the optimization of a quadratic function with multiple variables subject to linear constraints. It has prospects for use in model predictive control (MPC). The proposed successive overrelaxation (SOR)-like method utilizes a matrix-splitting scheme and can handle problems involving both state and input constraints. The corresponding algorithm can easily be implemented and can be tuned for optimum performance and global convergence. The method’s performance is analyzed and compared with those of other well-known and state-of-the-art methods via MATLAB simulations.