Optimization of Genetic Algorithm for Urban Traffic Light Schedule Problem

Abstract

Congestion is a major problem at urban traffic light intersections. One of the problems is the uneven distribution of vehicles at traffic light intersections, even though the portion of green duration at traffic light intersections is the same. Must present an adjusted deduction based on the number of vehicles at each intersection. In addition to optimizing the green duration, if you only consider one intersection, it will cause other problems such as congestion at the next connected intersection so that the required parameters are used for coordination between the initial intersection and the next connected intersection. In this research, optimization was carried out by giving weight to each parameter based on real vehicle data passing through the route at 11 intersections in Yogyakarta. At 11 intersections, 9 traffic light intersections were taken which will make the optimal green duration based on the parameters of the initial flow, destination flow and trip duration for traffic light coordination between the initial traffic light intersection and the next connected traffic light intersection. Next, the fitness function is formulated and then processed using a genetic algorithm. In the Genetic Algorithm optimization process, chromosomes are the green duration of 9 traffic light intersections which will then be processed with the genetic algorithm stages. The result is 9 optimal green durations based on the initial flow, the destination flow and the duration of the trip. The green duration adjusts the weights with maximum accuracy and reduces the vehicle travel duration from 44-64 seconds per one traffic light and a total of 419 seconds at 9 traffic intersections from the optimized data.