Discussão conceitual sobre métodos de reconstrução de matrizes origem-destino estáticas em redes de transportes

Abstract

Origin-destination (OD) matrices quantify the demand for transport in a geographic region, and play a key role in planning studies, design and management of urban and regional transport systems. Traditionally, the estimation of OD matrices consists of making direct sample surveys, such as: household interviews, vehicle plate counting and interviews in public transport terminals. A lower cost alternative is to synthesize an OD matrix by means of mathematical methods using traffic volumes observed in a transport network, which can be classified into methods of reconstruction and estimation. This paper presents a conceptual discussion focused on the assumptions and limitations of the main reconstruction methods based on maximizing entropy, generalized least squares and Bayesian inference. We describe in detail the mathematical foundation of the methods and make recommendations to their improvement and effective application