Abstract:

       Track quality index (TQI), a time series with random characteristics, is an important indicator reflecting the total unevenness of a track segment. We analyze the correlation between TQI sequence and every single indicator with grey theory and predict next inspection value of a certain segment. Moreover, we investigate TQI sequence data for 200 m track segment with autoregressive moving average (ARMA) model, analyze their variation tendency and predict TQI value in future time.Analysis show that the accuracy of each model has been improved, and the relative error is less than 5%.

Key words: time sequence, ARMA model, correlation, fitted GM(1,1), TQI