针对城市交叉口交通流的分布特点,给出了一种自适应交叉口多相位控制算法,考虑相邻车道上的车辆排队长度,利用多层BP神经网络实现了道路交叉口多相位模糊控制.仿真结果表明,文章所设计的模糊神经网络控制器能有效地减少单交叉口平均车辆延误,具有较强的学习和泛化能力,为实现交通系统智能控制提供了一条新途径. Aiming at the distribution characteristics of traffic flow in urban intersections, an adaptive multi-phase control algorithm for intersections is presented. Taking into account the queue length of vehicles on adjacent lanes, multi-layer BP neural network is used to realize multi-phase fuzzy control of road intersections. The simulation results show that the fuzzy neural network controller designed in this paper can effectively reduce the average vehicle delay at a single intersection and has strong learning and generalization ability, which provides a new way for intelligent control of traffic system.