Prediction of Railway Energy Consumption in Turkey Using Artificial Neural Networks

A number of measures are being taken to protect the rapidly depleted energy resources around the world. The trend towards sustainable energy resources is increasing, especially in order to improve energy efficiency in transportation vehicles. In this study, the total energy consumption amounts of the railway vehicles were examined based on the line length, number of passengers and the amount of cargo in the last 43 years in our country. For 5 different models created by the artificial neural networks method, the amount of consumed energy and estimated energy amounts were compared using the correlation coefficients, R-2, absolute error and absolute relative error criteria using Levenberg-Marquardt and Conjugate Gradient Descent algorithms. In the model 3, where the number of passengers and the amount of cargo were used as inputs, accuracy values and error criteria were better. According to the results obtained in the study, it was revealed that the amount of energy consumption is mostly related to the amount of load and then the number of passengers, and the change in line length and years is less effective. With the data obtained in this study, it will be possible to determine the amount of energy that can be spent by using the number of passengers planned to be transported on the railways in the future periods and the amount of cargo. Thanks to the determined amount of energy, a significant amount of savings can be achieved by focusing on sustainable energy resources.