平面交叉路口的神经网络自学习控制方案

针对城市道路平面交叉路口的交通信号控制，构造了一种具有实时学习功能的神经 网络智能信号控制方案．方案中的信号控制器由底层的两个神经网络和顶层的评价准则组成 ．两个神经网络总是交替处于学习和工作状态，评价准则则根据路口的车流情况确定是否需 要对神经网络进行训练．仿真实验结果表明，该控制方案能很好地适应路口的实际交通状况 ，从而达到有效提高路口通行能力的目的．

AN APPROACH OF INTERSECTION TRAFFIC SIGNAL CONTROL

This paper discusses a method of intersection traffic signal control with online self learning ability. The control system is composed of intersection, two neural networks and a performance evaluation unit. Two neural networks are always alternatively in the state of learning or working during the process of self learning according to the decision of the performance evaluation unit on the traffic conditions of intersection. Simulation results reveal that the proposed approach can better fit the actual traffic conditions than old ones, and thus it has reached at the goals of making good use of intersections.