A Novel Approach to PEMFCs Degradation Prediction: An Enhanced Refinement of the Gray Verhulst Model

In pursuing superior operational efficiency of proton exchange membrane fuel cells (PEMFCs), the precise prediction of lifespan becomes a prerequisite. To address the issues of traditional static indicators such as voltage and power, this article propounds the relative voltage loss rate (RVLR) as an innovative degradation indicator in dynamic operation contexts. An advanced refinement of the gray Verhulst model (GVM) is proposed for the prediction of PEMFCs degradation. First, to deal with the volatility of PEMFCs’ voltage leading to the roughness in PEMFs’ degradation indicators, a novel exponential hyperbolic sine function transformation is applied to enhance the smoothness of the PEMFCs’ indicator sequences. In addition, the approach uses Newton-Cotes and Newton interpolation techniques to reduce the original model’s background value error. Empirical validation is provided through three case studies involving aging tests from the YK-S20, RG-FCTS-15, and G20 test stations. The proposed method demonstrates substantial accuracy and robustness, evidenced by RMSE values of 0.0031, 0.0017, and 0.0012, $R^{2}$ values of 0.9851, 0.9984, and 0.9939, and MAE values of 0.0019, 0.0010, and 0.0007 across the three datasets, respectively. These results indicate the method’s promise for accurate degradation prognostication.