Constrained Model Predictive Control of PEM Fuel Cell with Guaranteed Stability

In this paper, a constrained model predictive controller with guaranteed stability is proposed for a PEM fuel cell. The aim is to prevent oxygen starvation by controlling the air supply system, when the control system is affected by required stack current as a measurable disturbance. The compressor voltage is controlled to regulate the oxygen excess ratio towards a desired equilibrium to avoid oxygen starvation. A dual-mode controller is utilized to guarantee input-to-state stability. In a neighborhood of the target state, the control action is generated by a local state feedback controller and outside this neighborhood model predictive control is employed. Linear Matrix Inequalities is used to obtain a terminal cost and a local state-feedback control law to satisfy MPC input-to-state stabilization conditions. A nonlinear dynamical model of PEM fuel cell is used as a simulator. Simulation results indicate that the proposed controller leads to improved stability and much less computations with respect to conventional GPC controllers.