Acessibilidade e mobilidade na estimativa de um índice de potencial de viagens utilizando redes neurais artificiais e sistemas de informações geográficas. / Acessibility and mobility in the estimation of a trip potential index using artificial neural networks and geographic information systems

AUTOR(ES)
DATA DE PUBLICAÇÃO

2000

RESUMO

In general, the transportation planning process is not sensible enough to solve or at least to reduce the gap between what is planned and the real needs of urban citizens, specially those who belong to low income classes. Besides, although the analyses often take into account accessibility elements and mobility components they rarely do it in an integrated manner. As an answer to this deficiency, the objective of this work is develop a modeling approach for estimating potential trips that integrates both aspects for strategic planning purposes. Based on a comprehensive literature review, a new methodology is then proposed. It starts with the integration of origin-destination (O-D) survey data and spatial data obtained in a Geographical Information System (GIS) environment. Next, a mean separation accessibility index estimated for all households must be linked to their mobility variables, such as income, for example, in the same database. The output variables, i.e. trip characteristics (number and length), can be taken from the O-D survey or calculated in a GIS-environment. Next, exploratory models should be built with Artificial Neural Networks in order to evaluate the behavior of input and output variables. Only those variables selected as the most relevant in the evaluation phase are used thereafter to rebuild the models and to generate the Trip Potential Index - TPI. The proposed approach has been tested in a case study carried out in a Brazilian medium-sized city. For the most part, the results obtained with the trip potential model here developed suggest its superiority when compared to a conventional, selfstanding accessibility measure for strategic planning purposes. An analysis of two correlation coefficients, the first one got when the model estimates are compared with the real trip values (r = 0.60) and the second one got when the model estimates are compared with the accessibility values (r = 0.21), also strengthen the previous statement. Size and income of the household, which may be associated to mobility, and the accessibility indicator itself, were selected as the most relevant variables in the model. The selection of those variables stressed the assumption that accessibility and mobility should be examined together in transportation planning analyses. In conclusion, for the level of strategic planning, the methodology presented in this work seems to be a step forward in relation to traditional accessibility models and a useful tool for urban and transportation planners and decision-makers. The approach makes clear that urban citizens need not only physical accessibility, but also better mobility conditions.

ASSUNTO(S)

sistemas de informações geográficas acessibilidade geographic information systems redes neurais artificiais artificial neural networks trip potential index accessibility índice de potencial de viagens mobility mobilidade

ACESSO AO ARTIGO

Documentos Relacionados