Data-Driven Fault Detection for SOFC system based on Random Forest and SVM

Accurately detecting faults is essential for the development and maintenance of safe and reliable Solid Oxide Fuel Cell (SOFC) system. In this paper, A fault detection framework combining data generation and data-driven modeling for SOFC system is proposed: The most important features are selected based on Random Forest and the Support Vector Machine with radial basis kernel function (rbf-SVM) is applied to realize multi-fault classification. The experimental results demonstrate that the method is feasible and effective, significantly improving the performance of fault detection and isolation (FDI).