In order to improve the accuracy of railway freight volume prediction, we adopted the BP neural network method combined with grey relational analysis (GRA). Given that BP neural network algorithm are prone to local minimum, slow learning convergence, and diversity of structural selection problems, we especially introduce GRA to optimize the prediction process. In the construction of GRA-BP prediction model, based on the statistical yearbook data, we first selected the key indicators such as railway freight turnover rate, raw coal output, railway operating mileage, highway and waterway freight volume and the added value of the primary industry as the main factors affecting the railway freight volume. These indicator data were then used as input for the GRA-BP model, where the first 70% of the data were used for model training and the last 30% for testing. After training and testing, we obtained the prediction results of railway freight volume, and calculated the evaluation indexes such as MSE, RMSE, MAE, MAPE and R2. The results showed that the GRA-BP prediction model performed well in the nonlinear fitting, and the prediction accuracy achieved the expected effect.
Railway Freight Volume Forecast Based on GRA-BP Model
Advances in Economics, Business and Management res.
International Conference on Management Science and Engineering Management ; 2024 ; Shenyang, China June 07, 2024 - June 09, 2024
Proceedings of the 2024 5th International Conference on Management Science and Engineering Management (ICMSEM 2024) ; Kapitel : 99 ; 990-998
22.11.2024
9 pages
Aufsatz/Kapitel (Buch)
Elektronische Ressource
Englisch
Railway Freight Volume Forecast Based on GRA-WD-WNN
IEEE | 2019
|Railway freight volume forecast using an ensemble model with optimised deep belief network
Wiley | 2018
|Railway freight volume forecast using an ensemble model with optimised deep belief network
IET | 2018
|