基于MP与MPC相结合的分布式交通信号控制研究

依托于最大压(Max Pressure，MP)分布式信号控制特性，结合已建立的实时排队长度预测模型，利用模型预测控制(Model Predictive Control，MPC)进一步对修正后的MP信号控制策略进行滚动反馈优化，提出了MP与MPC相结合的分布式信号控制方法. 模型验证结果表明：MPC嵌入后，MP优化下的交叉口1 与交叉口2 延误分别降低了13.47%、15.35%；并且对比了MP绿信比分配权重修正前后对控制输出的影响，优化后瓶颈路段的车道4 和车道10 的排队溢出次数分别从6 次和9 次降低为0 次，表明修正后MP绿信比权重的信号优化策略更能有效防止排队溢出现象的发生.

Distributed Traffic Signal Control Based on Combination of MP and MPC

Based on the distributed signal control characteristics of the max pressure (MP) and the real- time queue length prediction model established by the author, the model predictive control (MPC) is used to further optimize the rolling feedback of the modified max pressure signal control strategy. A distributed signal control method combining the MP and the MPC is established. The results of the model verification show that the delay of the intersection 1 and the intersection 2 after the MPC embedding is reduced by 13.47% and 15.35%, respectively, and the effect of the MP on the control output before and after the split weight correction is compared. And the number of queue overflow times of the lane 4 and the lane 10 of the bottleneck section after the optimization is reduced from 6 and 9 times respectively to 0. It is shown that the modified signal optimization strategy of the MP split can effectively prevent the occurrence of queue overflow.