W. Chaker, B. Moulin, and M. Thriault (Canada)
Multiscale modeling, hybrid traffic simulation, network representation, synthetic population, design pattern, travel demand modeling.
Being able to vary the level of detail or scale when modeling any system has an increasing interest in different domains. Here we address the issue of multiscale modeling of transport networks in order to enhance feasibility of hybrid simulations, like those who couple macro and micro traffic behaviours, or those who recently tried to combine cellular automata with multi-agent systems in urban simulation and geosimulations. Using an example, we show how a generic link/node representation which forms the core of a design pattern, can be used to instantiate several network models at different scales. Each one can be simulated using the appropriate behavioural model. The design pattern approach avoids drawbacks of strictly hierarchical representations and maintains coherency. We use a multi level spatial grid to locate vertices that form a link. This hierarchical grid is also a way to deal with behavioural models based on cellular automata. The concept of Place is introduced in order to be able to connect generated synthetic populations to the transport network and, then, to model the travel demand. Multimodality is allowed and opportunities of modal transfers are explicitly defined. The paper also shows how we are using real GIS data of Quebec City to build a three-scale transport network with the suggested approach.
Important Links:
Go Back
