Prediction of Track Quality Index Based on Combined Model

• 1、School of Computer Science, Beijing University of Posts and Telecommunications, Beijing 100876

Abstract：The Track Quality Index(TQI) reflects the degree of dispersion of the geometrical irregularity of the track in the unit section. It is the most important indicator for the maintenance and repair of the line by the railway engineering department of China. In order to accurately predict the development trend of TQI, this paper proposes a TQI prediction method based on non-equal interval weighted grey model and recurrent neural network. Firstly, the overall trend of TQI is predicted using the optimized non-equal interval weighted GM(1,1) model. Then, the randomness of TQI changes is studied and predicted using Recurrent Neural Network (RNN). Finally, the sum of the two predictions is used as the short-term prediction of TQI changes. The experimental results show that: 1) The effect of individual TQI value fluctuations on prediction accuracy of the optimized gray model is significantly reduced;2) Compared with the existing prediction models, the prediction accuracy proposed in this paper there has been a significant improvement.

Keywords： Technology of Computer Application Track Quality Index Grey Model Recurrent Neural Network