Aiming at the prediction of the development of rail corrugation, in-depth analysis was carried out based on actual field-measured rail wave data, and a combined model of rail corrugation based on non-equal interval grey model and BP neural network was proposed. The advantage of this model is that the gray theory requires less data, and the BP neural network has strong nonlinear fitting ability. It can predict the future development of rails based on the original wave depth values measured by a small number of non-equal time intervals. The historical data of rail wave wear depth on a certain line are used for model training and prediction analysis. The results show that The average absolute error of the combined model prediction results is significantly reduced compared to the single gray model, and the prediction accuracy test level reaches level 1. It proves the effectiveness of this prediction method in the prediction of rail grinding, which provides important guiding significance for the development of track maintenance and polishing strategy by the public works department.
Prediction of Rail Corrugation Based on Non-Equal Interval Grey Model and BP Neural Network
13th Asia Pacific Transportation Development Conference ; 2020 ; Shanghai, China (Conference Cancelled)
2020-06-29
Conference paper
Electronic Resource
English
Engineering Index Backfile | 1913
|Engineering Index Backfile | 1907
|Engineering Index Backfile | 1923
|Engineering Index Backfile | 1930