Train Bearing Temperature Trediction Based on LSTM Recurrent Neural Network

Abstract

The bearing temperature is an important indicator of the bearing state during the operation of the EMU train, and it is also an important physical parameter for the performance evaluation of bearing during long-term service. Due to factors such as train operating conditions, ambient factors, and thermal accumulation effects, it is difficult to obtain good results only by setting the bearing temperature threshold to evaluate the bearing operating state. Based on the big data application development platform, based on the analysis of a large number of operation data of bearing temperature, the LSTM recurrent neural network is used to establish the bearing temperature prediction model, and the EMU bearing operation data is used to train and test the model to obtain the full operating condition and the full climate environment. The model is verified by the actual operating datum of a certain EMU train. The comparison with the operational test data shows the correctness and effectiveness of the proposed method.