Track Irregularity Prediction Based on DWT-DLSTM Model

The long-term operation of high-speed railway will lead to track irregularity that will cause random vibration of the track system and affect driving safety. The accurate prediction of track irregularity is of great significance to the quality of high-speed railway. In this paper, we proposed a DWT-DLSTM model to predict the track irregularity for high-speed railway. Firstly, the track irregularity time series data is denoised through the discrete wavelet transform (DWT). Then the deep long short-term memory (DLSTM) neural network is adopted to predict the denoised data. Finally, the experiment results show that the proposed DWT-DLSTM model outperforms other traditional models and obtain more accurate prediction results for track irregularity.