Modeling historical traffic data using artificial neural networks

The Design-Hour Volume (DHV), which is defined as the 30th highest hour volume in a year, is a significant concept in transportation engineering and planning. Finding the DHV requires hourly traffic counts for an entire year. However, this becomes a challengeable task when part of the data is not collected because of different reasons, such as construction activities or hardware failure. In this paper, an Artificial Neural Network (ANN) approach is used to develop a DHV prediction model based on historical traffic counts. The model takes into account the correlation between DHV and other variables such as AADT, functional classification, and number of lanes. Results show that the ANN model is capable of providing accurate and reliable DHV estimates.