Diagnosis Method for Solid Oxide Fuel Cell Using Temporal Convolutional Neural Networks and Support Vector Machines

Fuel depletion is one of the critical factors affecting the reliable operation of Solid Oxide Fuel Cell (SOFC). To address the issue of identifying weak faults in the system, where initial characteristics are not obvious and fault type determination is challenging, a fault diagnosis method combining Temporal Convolutional Network (TCN) and Support Vector Machine (SVM) is proposed. First, TCN is used to extract temporal features from the fault data. These features are then input into the SVM for classification. This approach not only improves the accuracy of fault identification but also effectively reduces the computational load of the model. Experimental results show that the proposed TCN-SVM model outperforms the bayesian artificial neural network with feature selection (BANN-FS) model in SOFC fault diagnosis, achieving fault diagnosis accuracies of 91.2%, 99.7%, and 99.1% for slight degradation, moderate degradation, and severe degradation of the SOFC stack, respectively.