Hydrogen Leakage Fault Diagnosis in PEMFC Systems Based on Synchronous Heterogeneous Characteristics of Multiple Physical Fields

Conventional hydrogen leakage fault diagnosis methods for proton exchange membrane fuel cell systems based on a single physical field are characterised by insufficient sensitivity and a high false alarm rate. This is due to limitations in sensor response speed and the dynamic operating conditions of the fuel cell system. This paper proposes a hydrogen leakage fault diagnosis method based on synchronous heterogeneous features of multi-physical fields. This method aims to realise efficient extraction and correlation analysis of the coupled features of multi-physical fields. It is achieved by constructing an electric-thermal-gas multi-sensor collaborative monitoring network, combined with the Fusion Convolutional Transformer Neural Network. The experimental results demonstrate that the proposed method achieves a diagnostic accuracy of 99.24% for hydrogen leakage faults, outperforming single-physical-field approaches (which reach a maximum of 94.99%) by at least 4.25 percentage points. This enhancement significantly improves diagnostic robustness under complex operating conditions. This study provides a theoretical foundation for selecting hydrogen leakage detection technologies in fuel cell systems, offering significant reference value for enhancing the safety performance of such systems.