Uma metodologia de análise da operação do transporte público de Curitiba utilizando link streams

Abstract

Transport systems usually require transfer of passengers between trains, vehicles or buses, among others. Transfer times and transport capacity directly affect the quality of transport experienced by users. Moreover, the computation of performance indicators for complex transport systems involving several lines and connections may be a hard task. This thesis proposes a methodology for analysis of Curitiba collective transport operation using link streams. Link streams are temporal graphs which model relationships (links) made and unmade between objects (vertices) in time. By computing cliques of the corresponding link streams, it is possible to identify and quantify (numerically and temporally) nearby buses in a terminal. These cliques are computed by algorithms with good scalability. The main contribution is to use cliques for obtaining transport performance indicators as transfer time, bunching, congestion, and number of passengers transferred. The results are obtained for real-world data of Curitiba city in Brazil. They show transfer time distributions, rush hours for bus congestion, and estimates of number of passengers transferred based on maximal bus capacity. Although results are obtained for Curitiba, the proposed methodology can be extended to other transport systems involving transfer of passengers or freight. The proposed performance analysis can be used for supporting urban planners to improve collective transport operation.