Prediction of Gate Valve Closing Fault Trend Based on Relevance Vector Machine

The failure of the gate valve under the switching process is the main problem during the transmission of nuclear power facilities. This paper proposes a relevance vector machine based method to predict the fault of the gate valve. With the extraction of an envelope from the captured vibration data as the input feature, the relevance vector machine is trained with some groups of data and tested by the remaining. The result demonstrates that the proposed method can accurately predict the fault of the gate valve. This method has a guiding significance for the actual valve failure prediction.