Simulation of car-following decision using fuzzy neural networks system

Car-following is a kind of driving behavior which is difficult to be expressed precisely, and is affected by multiple factors. In the work here, fuzzy reasoning theory is utilized to establish a fuzzy inference system composed of multiple neural networks, which contains traditional back-propagation neural network and wavelet networks. Moreover, on the basis of using various information around a driver sufficiently, the fuzzy neural networks system successfully simulated car-following behavior in single lane without lane-changing behavior and generated reliable simulation results.