ESTIMATION OF PCU FACTORS ON URBAN - STREETS USING ANN

Abstract

The vehicular traffic in India is highly heterogeneous which comprises of vehicles like two wheeler, three wheeler, small car, big car, light commercial vehicle, truck busses etc. having different static and dynamic characteristics. As such vehicles tend to occupy any lateral position on the road depending upon the available road space at any given instant of time. 1-lence, the volume of such heterogeneous traffic is measured by converting different types of vehicles into its equivalent passenger cars and by expressing the volume in PCU (passenger car units) per hour. Interaction study of such vehicles moving in a heterogeneous flow is very complex. In this study, an attempt has been made to develop speed models for different vehicle categories on urban streets using Artificial Neural Networks instead of commonly used regression analysis. The speed models thus developed were further used to study the variations of PCU factors with traffic volume and carriageway width. The study shows that traffic volume has a decreasing effect on individual vehicular speed, though this effect is not same for all, reason being the variation in physical dimension and maneuvering efficiency of individual vehicles. The study also outlines that with increase traffic volume, PCU values for three wheelers, big cars and heavy vehicles increases and that of two wheelers decreases. Also, it has been found that with increase in carriageway width, PCU values of individual vehicles increases.