Study on the mid & long term prediction model of track geometry state based on the grey time-varying parameters theory

According to the geometrical irregularity development characteristics of tracks and on the basis of the grey prediction theory, the unequal time-interval grey model with time-varying parameters was established, where track geometrical irregularities test data were taken as the time series and the background values were optimized. Periodic functions are introduced to modify the model by residual analysis so as to better describe the complex function relationship among the influencing parameters and improve the fitting and prediction precision of the model. Using this model, track quality index (TQI) data were predicted and their accuracy was tested. The results indicate that the model can well reflect the time-dependent random fluctuation characteristics of TQI and realize high prediction and fitting precision. The model proves suitable for middle and long-term prediction, thus providing a new method to know well the law of track quality development.