基于多路口预测与实时配时合作的交通控制系统设计

交通流量的预测是实现智能交通控制的核心问题。利用电子车牌识别法对车辆进行计数，根据数理统计原理筛选对决策路口影响最大的样本数据，建立NARX动态神经网络预测模型，以路网多路口交通流的时间序列数据进行训练，预测后一天同时段的车辆数,计算出两方向总体配时差值，再分配于早晚高峰时段，并进行实验得出理想现象；严重拥堵情况下启动实时配时与应急方案。选取上海市长宁区常年拥堵的金沙江路与中山北路交叉口为中心点，进行MATLAB仿真实验，所得预测数据与实际值比较差值较小，验证了以多路口数据预测单路口的创新设计能在实际交通中应用，且能与实时配时方案合作，缓解交通拥堵。

Traffic Control System Design Based on Multi-road Prediction and Real-time Timing Cooperation

The forecast of traffic flow is the core issue of intelligent traffic control.Using electronic license plate recognition method to count vehicles, according to the principle of mathematical statistics to filter the sample data that has the greatest impact on decision intersections, establish a NARX dynamic neural network prediction model,training the time series data of road traffic at multiple intersections to predict the number of vehicles in the same period of the next day. Calculate the total time difference between the two directions and redistribute them in the morning and evening rush hou,and experiment to get the ideal phenomenonr;Real-time timing and emergency plans are initiated under severe congestion.Selecting the intersection of Jinsha River Road and Zhongshan North Road as the center,which are always congested in Changning District of Shanghai, using MATLAB to simulate, and the difference between the predicted data and the actual value was smaller.It verifies that the innovative design of a single intersection predicted by multi-road data can be applied in actual traffic and can cooperate with real-time time-planning solutions to alleviate traffic congestion.