Fast Constrained Generalized Predictive Control with ADMM Embedded in an FPGA

Abstract

Constrained model predictive control (MPC) usually requires the computation of a quadratic programming problem (QP) at each sampling instant. This is computationally expensive and becomes a limitation to embed and use MPC in plants with fast sampling rates. Several special solvers for MPC problems have been proposed in the last years, but most of them focus on state-space formulations, which are very popular in academia. This paper proposes a solution based on alternated direction method of multipliers, tailored for embedded systems and applied to generalized predictive control (GPC), which is a very popular formulation in industry. Implementations issues of parallel computation are discussed in order to accelerate the time required for the operations. The implementation in an FPGA proved to be quite fast, with the observed worst case execution time of 11,54 µs for the presented example. These results contribute to embed GPC applications in processes that are typically controlled by classical controllers because of their fast dynamics.