An NSGA-Ⅱ-based parameter tuning algorithm for EKF-based sliding mode controller of PEM fuel cells

This paper addresses the challenge of tracking an arbitrary power profile in a proton exchange membrane fuel cell (PEMFC) in the presence of measurement noise and disturbances. To this end, we used an extended Kalman filter (EKF) to estimate the internal states of the PEMFC in conjunction with an adaptive sliding mode controller (SMC) that has been shown to reduce chatter. The model used by the controller captures the internal dynamics and nonlinearly, and is accurate within 0.1% of the high-fidelity model. We developed the conditions necessary for the stability of the proposed controller based on the Lyapunov stability theorem. We also developed a systematic multi-objective optimization methodology of the controller hyperparameters to simultaneously minimizing tracking error, controller-chatter, and controller input using the non-dominated sorting genetic algorithm Ⅱ (NSGA-Ⅱ). The controller performance was demonstrated using multiple simulated experiments. Based on experimental results on desired signal data, we concluded that the proposed controller scheme can track desired power profiles within a 1% error.