Highway traffic density control based on RBF neural network

In this paper, we employ a macroscopic traffic model and RBF neural network to address highway density control problem. This control system is a nonlinear feedback closed-loop system. Firstly, a popular traffic model called the LWR (Lighthill-Whitham-Richards) model is formulated. Secondly, a direct RBF neural network control for discrete-time nonlinear systems is introduced. The control algorithm is also expounded. Thirdly, a traffic density controller based on the LWR model and the direct RBF neural network is designed. This approach is aimed at the nonlinear system, and there is no need to use linearization approximations. Finally, simulation results of the highway density control system are obtained by using the Matlab software. The control system has excellent density tracking performance even in an environment of interference signals, and the highway mainlines can achieve the desired traffic density. This nonlinear feedback control, as well as the RBF neural network control, offers a new means to highway ramp metering.