VARIABLE NEIGHBORHOOD SEARCH AND AGENT-BASED SIMULATION APPLIED TO OPTIMIZATION IN PUBLIC TRANSPORT NETWORKS

Various problems in urban mobility affect the population daily, compromising their quality of life. The urban public collective transport system (UPCT) can be used to minimize these problems. However, in Brazil, this system is frequently inefficient, with prolonged travel and waiting times. This article proposes a hub-and-spoke network structure to optimize the bus-based UPCT for a medium-sized city. A hierarchical agglomerative clustering and k-means sub-clustering are used in a constructive heuristic for the development of a transport network. The Variable Neighborhood Search metaheuristic showed a 22% reduction in the total distance of the network, and agent-based simulation evaluated the consequences of population displacement in the proposed network by assessing travel times.