Modeling operating speed using artificial computational intelligence on low-volume roads

Abstract

In recent years, Artificial-Computational Intelligence (ACI) have found increasing applications in management of transportation infrastructures. Examples of ACI applications can be found in highway management however, compared to transportation planning, research of ACI methods applied to infrastructure management has been relatively limited. In this study was used artificial intelligence ANN (Artificial Neural Network). In particular the objective of the research study is to compare the predicted operating speed on tangents and circular curves for low-volume roads by using two different statistical approaches.The starting point was to predict the operating speed on investigated tangents and circular curves elements by using four regression equations developed using a traditional ordinary least-squares method (OLS) as shown in a previous work of the authors. Then, the same database was used to calibrate new operating speed models by using ANN procedure. The results have shown that ANN models offer more reliable results in terms of predicted operating speed than those returned by OLS method on all circular curves and on tangents lengths greater than 500m. For tangents length less than 500 m, OLS method is to be preferred to ANN procedure.