Vehicle Delay Modeling at Signalized Intersections with Gene-Expression Programming

Abstract

The accurate determination of vehicle delays is crucial for effective intersection management. Because the optimization of signal duration in a signalized intersection is based on the reduction of delays, the rapid and accurate determination of the upcoming delay becomes an important step toward the solution. The delay times can be determined by field studies, but this approach is time-consuming and costly. An estimation of delays using analytical methods is also an approach used by transportation agencies. However, inaccurate predictions of this approach, especially in oversaturated traffic flows, is a significant disadvantage. In order to overcome these problems, estimation models based on artificial intelligence have been developed in recent years, and examples of their application are presented in the literature in which successful results have been achieved in the estimation of vehicle delays. Starting from this point, gene expression programming, an artificial intelligence technique, was used in this study to obtain models that can estimate vehicle delays rapidly and quite accurately. For the selection of this method, the high success rate of the gene expression programming technique for different engineering problems in the literature was crucial. In this study, three delay estimation models were developed by using different parameters. These models were tested with the data collected from 18 different signalized intersections in the Kayseri and Konya provinces in Turkey. When the results obtained were evaluated, it was seen that the models acquired using the gene expression programming technique were very successful in vehicle delay estimation.