Method of Turnout Fault Diagnosis Based on DBN-BiLSTM Model

Abstract

Turnout is one of the basic equipment in railway signal systems, which failure seriously affects the safety and efficiency of train operation. Taking the power curve of the S700K switch machine as an example, this paper presented a fault diagnosis method for railway switches based on deep belief network (DBN) combined with bidirectional long short-term memory network (BiLSTM). First, the feature extraction of the original data is achieved by unsupervised training deep belief network; then the extracted features are used as inputs to the bidirectional long short-term memory network to realize the fault diagnosis of the turnout. BiLSTM has certain advantages over other methods in dealing with time series problems. Finally, the model is verified, which can effectively improve the accuracy of turnout fault diagnosis.