Application of GRU in Prediction of Subway Wheel Wear

Abstract

Predicting various indicators of the wheels can help the subway operation and maintenance department to formulate a reasonable maintenance plan, allocate maintenance resources, reduce maintenance time costs, and improve maintenance efficiency. In this paper, based on the analysis of the current research of wheel wear prediction at home and abroad, a tread wear sequence prediction model combining Ensemble Empirical Mode Decomposition (EEMD) and Gate Recurrent Unit (GRU) modeling method is proposed to address the time lag problem in the existing prediction. Firstly, EEMD algorithm is used to decompose the tread wear sequence into different intrinsic mode function(IMF), then the neural network is trained and predicted for each component, and finally, the prediction results of each component are linearly superimposed to obtain the prediction results of tread wear. In order to illustrate the effectiveness and superiority of the algorithm, a variety of prediction methods were applied and analyzed based on the actual measured train wheel wear data. The results show that after EEMD processing, forecast accuracy has been improved.