General Regression Neural Network Method for Delay Modeling in Dynamic Network Loading

Abstract

In vehicular traffic modeling, the effect of link capacity on travel times is generally specified through a delay function. In this paper the Generalized Regression Neural Network (GRNN) method that supports a dynamic network loading (DNL) model is utilized to model delays on an unsignalized highway node. The presented DNL model is constructed with a linear travel time function for link performances and an algorithm written with a set of rules considering the constraints of link dynamics, flow conservation, flow propagation, and boundary conditions. The GRNN method is utilized in the integrated model structure in order to provide a closer functional approximation to pre-defined flow-rate delay function, a conical delay function (CDF). Delays forming as a result of capacity constraint and flow conflicting at an unsignalised node are calculated with selected GRNN configuration after calibrating the neural network component with the CDF formulation. The output of the model structure, run solely with the CDF, is then compared to evaluate the performance of the model supported with GRNN relatively.