Fault Prediction of Turnout Equipment Based on Double-layer Gated Recurrent Unit Neural Network

Turnout is an important railway equipment, and the failure of the turnout equipment will have a great impact on the safe operation of the train. In order to predict the failure of the turnout equipment in advance, this paper combines the double-layer gated recurrent unit (DL-GRU) neural network with the failure prediction of the turnout. This paper extracts the features of the current curves generated during multiple actions before the turnout fails, and uses the method of kernel principal component analysis (KPCA) to reduce the dimensions of the extracted features. Finally, the time series data set of turnout action current fault feature is established, which is used as the input of the DL-GRU neural network to realize the fault prediction of the turnout. The simulation results show that the DL-GRU network has a high prediction accuracy, compared with LSTM network and single-layer GRU neural network, the DL-GRU has better prediction performance.