Development of a Modified Fuzzy Logic Traffic Light Controller Using Fuzzy Time Series for Predicting Traffic Flow

Abstract

This research work developed a modified fuzzy logic traffic light controller using Fuzzy Time Series (FTS) to predict traffic flow for single intersection with the aim of minimizing the average waiting time of vehicles at the intersection or junction. The static nature of green light timing in traditional traffic light control systems may lead to high traffic congestion and results in a huge delay of traffic at intersection. The technique applies fuzzy time series which predicts the future state of the phase based on the previous state. This technique allows the fuzzy controller to make its decision taking into account not only the current traffic situation as detected by the Wireless Sensor Network, but also the probable short-term evolution of the traffic conditions. In this way, the choice of the phases depends on the number of vehicles in the queue, while the green time duration of the traffic lights would be determined based on the traffic flow forecast by the fuzzy time series. The performance of the modified fuzzy logic light controller using FTS is compared with the traffic lights dynamic control system that incorporate wireless sensor network for real-time traffic observing with multiple fuzzy logic controller, using average waiting time per phase as performance metrics. All are carried out in MATLAB 2015a. The performance of the developed controller was compared with the standard fuzzy logic controller and it was found that for the standard controller, the average waiting time in scenario I for phase I, II, III and IV were 18.71, 18.26, 26.10 and 27.32 respectively while that of the developed controller were 17.86, 17.39, 25.54 and 26.25 which is an average of 3.8% improvement across the four phases. For scenario II standard controller the average waiting time for phase I, II, III and IV were 20.36, 19.82, 19.48 and 18.25 while that of the developed controller were 19.99, 19.31, 18.79 and 17.80 respectively which has an average of 2.8% improvement across the four phases. The scenario III average waiting time for phase I, II, III and IV were 21.20, 22.46, 26.23 and 27.35 respectively, while that of the developed controller were 20.45, 19.99, 25.67, 25.97 respectively and have an average of 5.43% improvement across the four phases.

while that of the developed controller were 19.99,  19.31,  18.79  and  17.80  respectively

which has an average of 2.8% improvement across the four phases. The scenario III average waiting time for phase I, II, III and IV were 21.20, 22.46, 26.23 and 27.35

respectively, while that of the developed controller were 20.45, 19.99, 25.67, 25.97 respectively and have an average of 5.43% improvement across the four phases.