Mathematical Model For Optimizing The Active Control of Traffic Signals Parameters

The vehicle fleet circulating in Brazilians' city centers has increased significantly, but the mobility public policies have not been sufficient to follow this growth. The consequence of it is the increase of vehicular traffic on roads leading to congestion. Given this scenario, there are several works in the literature on the construction of smart cities by the use of Computational Intelligence techniques. Among the strategies used, stands out the reactive traffic signals. In this context, this paper presents a new mathematical model called: OACTS – Optimization of the Active Control of Traffic Signals. This model associates the use of Genetic Algorithm (GA) to optimize the following traffic signal parameters: green time, cycle time, phase sequence and offset. For this, the model uses a new proposal of delay time function to evaluate the solutions of the algorithm. For testing the applicability of OACTS, a case study is developed in which real vehicles demand data of a Belo Horizonte region - Brazil (actual data collected by the city transport company - BHTrans) are loaded in the AIMSUN traffic simulator. This simulator is calibrated previously to better represent the real conditions of the study region. The OACTS model receives these demands from AIMUSUN and, based on the application of GA, defines, in real time, a good set of the traffic parameters that meets the demand in order to reduce the delay of vehicles in the intersections. The experiments were statistically analyzed and the results show that the OACTS was able to find better solutions to the problem when compared to other models that optimize a smaller set of parameters and to the traffic plan currently used by BHTrans.