Localização de sensores de tráfego com limitações quanto ao número de equipamentos e de faixas observadas : técnicas heurísticas e um branch-and-cut

Abstract

The network sensors location problem to monitor the vehicles flow has been an object of increasing interest in recent years, due to its relevance to traffic management and control, by providing data for the development of new public policies related to the transport sector and direct government decisions. In this sense, this dissertation presents a mathematical model of sensors location for vehicle counting, installed in road segments, for the problem of vehicle flow estimation between source and destination pairs, which incorporates constraints regarding the number of sensors and lanes observed. In addition, this work proposes an exact algorithm, Branch-and-Cut, and three heuristics that can be incorporated into it to improve computational performance. Based on real data from the Brazilian road network, computational tests were runned to evaluate the behavior of Branch-and-Cut with and without heuristics. The results show that Branch-and-Cut with heuristics provides better solutions in 49.59% of the evaluated scenarios, versus 16.67% of pure Branch-and-Cut.