The centrifugal air compressor plays a critical role in ensuring the stable and efficient operation of fuel cell systems. This study focuses on the precise prediction of decay characteristics in fuel cell centrifugal air compressors. Initially, a 5000 h durability test was performed on a fuel cell air compressor. Based on the experimental data, an air compressor performance decay prediction model was established using the Category Boosting (CatBoost) algorithm. The model was optimized using the Gray Wolf Optimizer (GWO) to improve its prediction accuracy. Furthermore, the input features were ranked and analyzed using the SHapley Additive exPlanations (SHAP) method. The results indicate that with an increase in test duration, both the outlet flow rate and pressure of the air compressor exhibit varying degrees of attenuation. The performance decay rate of the air compressor first increases and then decreases. Notably, the developed prediction model is capable of accurately predicting the performance MAP of the centrifugal air compressor after attenuation, with the absolute percentage error controlled within 5%. This study presents an effective approach for predicting the degradation performance of fuel cell centrifugal air compressors, which is of great significance for the durability design and performance optimization of fuel cell air compressors.
Ma et al. (Sun,) studied this question.