Application of Elman Neural Networks to Predict Truck’s Operating Speed
Shao Bo Xie, Lang Wei · Advanced materials research · 2013
Prediction of vehicular operating speed is critical to evaluate the design consistency of road alignment. Elman neural networks are proposed to predict the truck’s 85th percentile operating speed. A total of 190 samples are collected from the two-lane rural roads and two factors are considered as input variables to the model including the curve radius and longitudinal slope. 100 samples are applied for training the networks to get the prediction model and the other 90 samples are used for the model validation. Additionally, the Elman neural networks are compared with back-propagation neural networks and linear regression, and the results show that the Elman neural networks are prior to the other two approaches and can be regarded as an alternative to predict truck’s operating speed.