ICTACT Journal on Soft Computing

Min Keng Tana ¹, Helen Sin Ee Chuo ¹, Kit Guan Lim ¹, Renee Ka Yin Chin ¹, Soo Siang Yang ¹, Kenneth Tze Kin Teo ¹

Affiliations
1 Modelling Simulation and Computing Laboratory, Artificial Intelligence Research Unit, Faculty of Engineering, Universiti Malaysia Sabah, MY

Abstract

Traffic congestion is a perennial issue for most cities. Various artificial intelligence (AI) algorithms, which can categorize as deterministic and metaheuristic algorithms have been suggested to mitigate congestion. Although traffic flow is dynamic and stochastic in nature, most of the previous works evaluated the algorithms with a deterministic or non-stochastic traffic flow pattern. As such, the adaptiveness of those AI algorithms in dealing with stochastic traffic flow patterns is yet to be investigated. Therefore, this paper aims to explore the feasibility of both algorithm types in controlling stochastic traffic flow. In this work, a benchmarked traffic flow model is modified and developed as the simulation platform with the parameters extracted from the guidelines of Public Works Department Malaysia (JKR). Normal distribution function is embedded in the developed model to simulate non-uniform headway for inflow and outflow vehicles. Two commonly used algorithms, namely Fuzzy Logic and Genetic Algorithm are proposed as the adaptive controller to optimize the traffic signalization based on the instant stochastic traffic demand. The simulation results show the metaheuristic algorithm performs better than the deterministic algorithm. The mutation mechanism of the metaheuristic algorithm improves the exploration ability of the algorithm in seeking the optimum solution within the solution space without bounded by a set of fixed-computational rules.

Keywords

Genetic Algorithm, Fuzzy Logic, Signal Optimization, Stochastic Flow, Saturated Condition.

Full Text