Enhancement of time of day based traffic signal control

A majority of urban traffic signal systems including coordinated actuated signals deploy multiple signal timing plans to account for traffic demand changes over time during a day (i.e., off peak, morning peak, mid day, evening peak, and night). A recent study proposed a data driven methodology that develops such time of day (TOD) break points using a clustering algorithm. One disadvantage of such statistical approaches was that they often produce clusters with few data points (i.e., infeasible time-of-day interval). This paper presents an enhancement of data driven methodology that determines optimal break points for TOD traffic signal control. In order to overcome such drawback, a genetic algorithm based clustering is proposed and evaluated through the performance of actual timing plan under microscopic simulation program, SIMTRAFFIC. The study results indicate that the proposed method can eliminate infeasible TOD intervals and its TOD-based timing plan produced comparable performances to those existing TOD intervals.