Optimization of Time-of-Day Breakpoints for Better Traffic Signal Control

Abstract

Traffic signal control is one of the most cost-effective means of improving urban mobility. Signal control can be categorized as pretimed, actuated, and adaptive. Among these, both pretimed and coordinated actuated controllers deploy multiple signal timing plans to account for traffic demand changes during the day, whereas adaptive control changes the timing plan in real time according to traffic conditions. In the case of pretimed and coordinated actuated signals, morning peak traffic would differ from that of the off-peak such that it would be better to use two distinctive signal timing plans. Traffic engineers often determine such time-of-day (TOD) breakpoints manually by using 1 or 2 days' worth of traffic data. A few recent studies developed statistical and heuristic methods for TOD breakpoints by using archived traffic data. These approaches determined the breakpoints through minimization of within-cluster distance and maximization of between-cluster distances. Thus, the clusters do not directly reflect the performance of timing plans and often result in only local optimal TOD breakpoints. One method is based on a genetic algorithm (GA) that optimizes TOD breakpoints with explicit consideration of signal timing performance at a representative intersection. The proposed method implements two-stage optimizations: outer loop for TOD breakpoints and inner loop for timing plans of corresponding intervals. The proposed approach is implemented on a network consisting of three coordinated actuated signalized intersections. The convergence graphs of both inner-and outer-loop optimization indicate that the GA-based algorithm obtains breakpoints within a relatively small number of iterations. Also studied was the performance of the proposed approach for a varying number of breakpoints (i.e., four to eight). The results, based on a microscopic simulation program, Sim Traffic, indicated that six breakpoints outperformed the other numbers of breakpoints considered.

A procedure for determining time-of-day break points for coordinated actuated traffic signal systems

“…Note that microscopic simulation is required to explicitly account for transition impacts. This is why previous studies (Smith et al, 2001;Park et al, 2004a;Wang et al, 2005 ) did not explicitly consider transition impacts during the optimization of TOD break points.…”

“…Initiated from the challenges of using a statistical clustering approach, Park et al proposed a heuristic clustering approach based on a Genetic Algorithm (GA) and developed an automation process to identify optimal break points (Park et al, 2004a). In their research, a GA-based optimizer was applied to find the optimal timing plans for each TOD period and explore the optimal TOD break points.…”

“…A few recent studies (Smith et al, 2001;Park et al, 2004a;Wang et al, 2005) have demonstrated that the selection of Timeof-Day (TOD) break points could significantly enhance day-today traffic operations. This idea has been adopted in the ACS Lite (Luyanda et al, 2003) algorithm.…”

Traffic predictive control from low-rank structure

“…Approaches for TOD segmentation proposed in the literature include randomized clustering algorithms [29,30], heuristic genetic algorithms [31], and simulation based algorithms [32]. To improve performance, the paper [30] suggests explicitly incorporating time as a variable in determining the clusters.…”

A genetic algorithm-based procedure for determining optimal time-of-day break points for coordinated actuated traffic signal systems

“…Park et al (2001) proposed a GA-based optimization technique to find the best TOD break points. Their proposed GA-based optimization consisted of two separate loops: the outer loop, which searched for the best TOD interval, and the inner loop, which optimized the best timing plans of each TOD interval generated by the outer loop.…”