摘要： 针对交通数据在传输过程中随机丢包造成交通拥堵的问题，提出一种新的交叉口排队长度均衡控制方法。考虑到交叉口交通控制的重复特性和强非线性，将无模型自适应迭代学习控制方案应用于交叉口排队长度控制中，通过实时调整各交叉口的信号配时方案来调节路口车辆的排队长度，实现各交叉口排队长度的均衡。针对道路交通网络控制中排队长度差值数据在传输过程中存在的丢包现象，将数据丢失现象描述为概率已知的伯努利序列，提出数据丢失情况下的补偿算法，即利用上次迭代的输出数据、伪梯度的估计值和控制输入差值对丢失数据进行补偿，解决存在数据丢包情况下多交叉口排队长度均衡控制问题。仿真结果表明，该方法在数据丢包的情况下迭代100次左右能够收敛于期望值并达到期望控制效果，验证了补偿算法的有效性。

关键词: 交通拥堵, 数据丢包, 无模型自适应迭代学习控制, 信号配时, 排队长度均衡, 补偿算法

Abstract: To address the traffic congestion caused by random packet dropout in the process of traffic data transmission,this paper proposes a method to control the queue length balance at intersections based on data packet dropout.Considering the repetitive characteristics and strong non-linearity of traffic control at intersections,the Model-Free Adaptive Iterative Learning Control(MFAILC) scheme is applied to the queue length control at intersections, adjusting the queue length of vehicles in the intersection by changing the signal timing scheme for each intersection in real time to achieve the balance of long queues at intersections.According to the phenomenon of packet dropout of the queue length difference data in the transmission process in the road traffic network control,the data dropout is described as a Bernoulli sequence with known probability.On this basis a compensation algorithm under data dropout is proposed, which compensates for the lost data by using the output data of the previous iteration,the estimated value of pseudo gradient and the control input difference to solve the problem of queue length balance control in the case of data packet dropout.Simulation results show that the proposed method can converge to the expected value and achieve the desired control effect after about 100 iterations in the case of data dropout,which verifies the effectiveness of the compensation algorithm.

Key words: traffic congestion, data packet dropout, Model-Free Adaptive Iterative Learning Control(MFAILC), signal timing, queue length balance, compensation algorithm

中图分类号: 

• TP391