Performance study of artificial neural network modelling to predict carried weight in the transportation system

The major aim of this study is to model and predict the amount of carried weight based on the five direct impact factors in the transportation system. In this study, artificial neural network (ANN) has been incorporated for developing a predictive model. Three different training algorithms, namely Levenberg-Marquardt-LM, batch backpropagation-BBP and quick propagation-QP, were used to train. The input parameters are the aforementioned five transportation factors plus two timing factors namely number of weeks and seasons while the carried weights is the output. The next purpose of this study is comparing the mentioned learning algorithm's performance based on predicting ability. The results showed that the QP algorithm with 7-4-1 network topology exhibited the highest predictive power. The available data have been trained by ANN (QP-7-4-1) and the responses were predicted. Moreover, the truck factor plays a slightly more dominant role in the prediction of carried weighs.