Özet:

One of the biggest problems of today is environmental pollutants and their effects. When the sources of environmental pollutants are examined, many sectors are under examination. One of these sectors is the transportation sector with 19% emission share in all sectors. When the modes of the transportation sector are examined, the most polluting is road transportation. Therefore, the general policy of the countries is to focus on rail and maritime transport, which saves fuel consumption in both freight and passenger transport. Along with saving in fuel consumption, CO2 emissions also decrease. In this study, change of railway transport mode share are examined for Turkey and prediction models are created by using artificial neural networks (ANN) method for rail transport CO2 emission. In the prediction models, six variables are examined: length of railways, the number of passengers, freight amount, train-kilometer, passenger-kilometer and ton-kilometer for railway transportation. The models are created according to the combinations of input variables with 2, 3, 4, 5 and 6 combinations. New inputs are created by using principal component analysis (PCA) in order to both increase performance and predict the effects of the six variables used in the model. As a result of the performance evaluation of the models, the best prediction model is NPFA which have 2 variables. Mean Square Error (MSE), Mean percentage error (MPE) and coefficient of correlation (R) performance values of the model are 6.48 × 10-5, 0.125% and 99.65%, respectively. In addition, PCA has significant effects on models.

Anahtar Kelimeler: