Modelling of the Interdependence Between Speed and Traffi c Flow Density. A Neuro – Fuzzy Logic Approach

The speed-traffi c fl ow density interdependence diagram has a number of varia ons, star ng with the theore cal model, through various empirical models that were developed and models based on actual research done on traffi c fl ow. The func onal interdependence is obtained using the Sugeno fuzzy logic system, where representa ve values proposed in HCM 2010 have been adopted as parameters of output associa on func ons. Subsequently the neural network is trained based on actual traffi c fl ow data, which by adjus ng the associa on func on of the fuzzy logic system yields an output form of the basic traffi c fl ow diagram. It was no ced that this hybrid expert system produces be er output results by applying the “subtrac ve clustering“ method on data that are used for training a neural network. Finally, the model was tested on several input data groups, and the interdependence between speed and traffi c fl ow density is shown in graphical form. Keyword. Basic traffi c fl ow diagram, traffi c fl ow theory, neural networks, fuzzy logic, subtrac ve clustering, hybrid expert systems.