A Feature-Based Prognostics Strategy for PEM Fuel Cell Operated under Dynamic Conditions

Proton exchange membrane (PEM) fuel cell has been widely used in diverse applications, especially in automotive field. However, durability and cost are two principle barriers to take PEM fuel cell into wide commercial use. The discipline of prognostics and health management (PHM) tends to help against this problem. PHM aims at deploying predictive maintenance and anticipating degradation mitigation strategies for PEM fuel cell. This paper focuses on the prognostics of PHM and an adapted data-driven fuel cell prognostics approach based on multiplicative feature decomposition and echo state network is proposed to predict the fuel cell degradation behaviour under dynamic operation conditions. Experimental data is used to verify the effectiveness of the proposed algorithm.