engleski

Application of neural networks for input parameters optimization in VISSIM

It is unquestionable that microsimulation models are a very useful tool for analysis of existing critical segments of a network, in this case of roundabouts, as well as prediction of traffic conditions at existing and planned intersections. However, the question is whether they can give expectedly realistic results of modelling in order to be applicable in methodology of analysis and planning of roundabouts in the local conditions. For a successful application of microsimulation models including VISSIM in the local conditions it is obligatory to perform calibration and verification of the models. Input parameters optimisation of the models when performing calibration is customary world praxis, but the methodology of calibration and optimisation is still in the testing phase and hasn’t been adopted jet. This paper is a result of a research of the possibilities of neural networks application in input parameters optimisation procedures of the VISSIM model.

microsimulation; VISSIM; calibration; neural networks

nije evidentirano