Performance improvement of an NMPC problem by search space reduction and experimental validation to a PEM fuel cell system

Abstract

The current work addresses the control issues that arise during the operation of a fuel cell system based on a novel combination of two Model Predictive Control strategies, explicit and Nonlinear MPC (NMPC). The proposed framework relies on an NMPC formulation that uses a simultaneous direct transcription dynamic optimization method that recasts the multivariable control problem into a nonlinear programming problem using a warm-start initialization method. The performance of the optimizer is improved by a search space reduction technique which is based on a piecewise affine approximation of the variable's feasible space, derived offline by a multi-parametric Quadratic Programming method. The behavior of the explicit NMPC framework is initially explored by a simulation study and subsequently it is experimentally verified through the online deployment to the fuel cell unit, demonstrating excellent response in terms of computational effort and accuracy with respect to the control objectives.