Modelling of Delay for Protected/Permitted Left Turning Vehicles using Multigene Genetic Programming

Abstract

Vehicular delay represents one of the fundamental traffic signal performance measures. In the past, number of delay models were developed mainly to estimate delays for exclusive phases (movements). In cases of left-turn movements that are served in protected/ permissive mode, there are very few models that can be used to estimate delays. Previously developed models for left/turn protected/permissive mode are based on a number of assumptions related to vehicular arrival/departure patterns. Thus, when used to estimate delays in a real-time manner, such models are prone to erroneous estimates. In this study, to overcome the limitations of current models, authors proposed a novel delay model for protected/permitted left turn operations based on Multigene Genetic Programming (MGGP) technique. Relevant data were collected on a cycle-by-cycle basis using the microsimulation model of real-world arterial. Using the MGGP, a novel delay model and its analytical formulation were proposed and compared with the benchmark model from the literature. The results indicate that the proposed model is more accurate and reliable and should be used as an alternative to traditional models. To strengthen the conclusions of our study, future work is mainly related to the expansion of the utilized dataset used for model development based on the field data. It is imagined that such an expanded dataset and additional options within MGGP will be explored to develop a more robust delay model.