Effect of Monitoring System Structure on Short-term Prediction of Highway Travel Time

This article discusses how the structure of the measurement system affects the short-term forecasts of travel time based on it. The effects of section length and location of different measurement stations were investigated. The study used empirical data. The research was carried out on a 28-km long interurban two-lane highway section. The prediction models were made as feedforward multilayer perceptron neural networks. The main results showed that the division of long road sections into shorter sub-links in the travel time measurement system was important. Furthermore, it would be crucial to obtain information about traffic flow rates entering the section in order to time the start of congestion correctly. In conclusion, the structure of the monitoring system should be based on the analysis of a typical location and the development of congestion along the section.