Design Of Intelligent Traffic Control System Based On Neural Network

The efficiency of Urban Traffic Control System for traffic flow scheduling will directly affect the traffic status, so an efficient traffic control system is in need. After investigations, the mode of fixed time exists in most of the scheduling strategy for traffic lights, but this mode cannot work well in the period of traffic flow in quite unstable status. Thus, research of traffic control system of real-time setting is significant, which is based on the traffic flow amount. In this article, the method of ITCS based on neural network is proposed to deal with the traffic status of randomness, complexity and uncertainty.Artificial neural networks is capable of self-learning and self-adapting. It is used for the system which cannot establish an accurate mathematical model. According to the mechanism of crossroads, I set up the BP neural network model with four input neurons cell and four output neurons cell. The BP neural network can predict the passing time needed after training of the traffic elements in four directions. Simulated by MATLAB, it proves that the model is able to predict the passing time needed for the queues of vehicles.This system is composed of hardware modules, DSP image processing module and the software of remote upper computer. The hardware circuit consists of one control module, driving modules, yellow flash module, etc. Control module is the core of the control system, responsible for the coordination of each function module. The task of driving module is to control the traffic lights and detect the fault. The yellow flash module is made of pure hardware, launching warning by yellow signal in fixed cycle when machine fault appears. The DSP image processing module can calculate out the traffic amount from the traffic image, and predict the impending traffic status via the neural network. The upper computer plays the role in the remote monitoring.The algorithm is realized by the DSP processor and the network model is trained with the sample of moving vehicles. After incessant regulation with the weights and thresholds, the simulation result finally met the requirement. The CAN bus is a communication bus, which could improve the reliability and real-time of the system. Anti-interference module is embedded to the system for protection. Consequently, this design has provided a new direction for the development of traffic controller, which is able to improve the efficiency of traffic scheduling.