Track Quality Index and Its Sub Items Prediction Based on Markov Residual Corrected Grey Model
To study the trend of changes in Track Quality Index(TQI)and its sub item values,this paper considered the unequal time interval characteristics of motion detection data and constructed a first-order grey model about TQI and its sub item values.After the original data was accumulated and transformed,the grey parameters of the model were solved using the least squares method,and the Markov residual correction method was introduced to correct the grey model.By comparing the output values(fitted value and predicted value)of the grey model TQI and each sub item before and after correction with the measured values,the improvement effect of residual correction on reducing errors and improving prediction accuracy was verified.The results show that after Markov residual correction,the grey model TQI and its sub item output values can better reflect its fluctuation situation.Compared with the pre-corrected grey model,the average error between the predicted TQI value and the measured value of the corrected grey model is decreased from 1.59%to 0.89%,and the overall prediction accuracy is improved from level 2 to level 1.The average error between the predicted value and measured value of each TQI sub item has been decreased from 3.18%to 1.42%,and the overall accuracy has been improved from level 2～4 to level 1～2.From the perspectives of reducing errors and improving prediction accuracy,the prediction performance of the corrected grey model has been significantly improved.
railway trackTQI predictiontheoretical analysisTQITQI sub itemgray modelunequal time intervalmarkov residual corrections
NETL
NSTL
万方数据
