A Temporal Convolutional Network Based Method for Fault Diagnosis of DEH System

The Digital Electric Hydraulic (DEH) system is quite important for the safe operation of steam turbine, and the fault diagnosis of DEH has attracted a lot attention for the accident reduction of turbines. In this paper, a Temporal Convolutional Network (TCN) based fault diagnosis method was proposed to monitor the running state of DEH system. Building the offline modeling on the data generated from a simulated DEH system, the fault diagnosis performance was evaluated on independent data. Comparing with other methods, such as LSTM and SVM, our TCN model outperformed them in both fault detection accuracy and false alarm rate, which indicated that our proposed TCN model had great potentiality in the application of DEH system.