Modeling of fuel cell operating condition estimation based on transfer learning

Proton exchange membrane fuel cell has become the most widely used fuel cell type in fuel cell vehicles. Estimation of fuel cell operating conditions can ensure the safety of the system, determine whether the system is faulty, and provide help for controller design. Aiming at the air supply system of PEMFC, considering the non-linearity and coupling of the system, a condition estimation algorithm based on the TrAdaboost algorithm is proposed to estimate whether the membrane thickness of the proton exchange membrane inside the stack is normal or not. Through the analysis of the model, the key variables affecting the membrane thickness of the stack are written into the system measurement values, and the working conditions of the stack are estimated by using the proposed algorithm. Finally, the proposed hybrid algorithm is verified on the Matlab simulation platform, and the results show that the proposed method can accurately track the real-time working conditions of the fuel cell stack.