基于动态可靠性反馈的城市应急疏散交通流主动控制研究

This research is for the needs of urban emergency planning. Focus on what kind of proactive control is effective for network paralysis prevention in presence of short-term high traffic demands and disturbance uncertainty. The goal is to provide emergency decision method for urban transportation system.. Heavy evacuation demands accumulated within a short time tend to cause oversaturated network. Uncertainty disturbances increase the fluctuation of oversaturation momentum, which may lead to random failure of control performance. Build oversaturation control-based dynamic reliability to estimate the generation probability of this random failure, and be used as performance index for evacuation capacity of the network control strategy, and even serve as a feedback index in the proposed loop optimization methodology presented by “centralized control, performance feedback and localized process” for urban emergency evacuation. First, Construct centralized evacuation traffic control model with no constraint on oversaturation control, for the consideration of solvability. Second, depending on the feedback of oversaturation control-based dynamic reliability in the optimized evacuation traffic, identify and relieve the oversaturated intersection group and bottleneck-causation traffic demand. Third, based on the requirement of traffic state improvement, explore the technique of fixing partial traffic signal parameters which would return back to centralized control. This proactive control could reduce the total evacuation time and prevent oversaturated emergency evacuation traffic simultaneously. Develop simulation test and conduct empirical study in engineering project featured with varied emergency events and network structures.

面向城市突发事件应急预案编制需要，考虑疏散交通流时短量大特征和普遍存在的不确定性扰动影响，回答采取怎样的主动控制方法去应对潜在路网瘫痪问题，为城市交通系统的应急决策提供方法论。. 短时内大量积聚的疏散需求容易导致路网过饱和，不确定性扰动进一步加剧过饱和态势的波动性，易使控制性能随机失效。建立过饱和控制的动态可靠性刻画失效概率，作为体现路网控制方案疏散能力的性能指标，并用于构建“集中控制，性能反馈，局部梳理”的循环至优方法体系。先出于求解算法和运算成本的可实现性考虑，构建无过饱和控制约束的集中式疏散交通信号优化模型；再根据优化后的交通流及动态可靠性指标反馈，设计过饱和交叉口群和致堵需求源的甄别治理技术；然后面向过饱和程度逐步降低的循环收敛要求，探索用于返回集中式控制的信号参数的固定策略；最终达成总疏散时间优化和过饱和预防双重目的。以多类突发事件和相异路网结构为例开展仿真测试与实证研究。

建立过饱和控制的动态可靠性刻画失效概率，作为体现路网控制方案疏散能力的性能指标，并用于构建“集中控制，性能反馈，局部梳理”的循环至优方法体系。此动态可靠性指标与常规出行时间可靠性进行对比，在疏散路网中测试发现动态可靠性指标虽然不能减少过饱和路段，但是可以减少过饱和离散时段数（5min）达26%。先出于求解算法和运算成本的可实现性考虑，构建无过饱和控制约束的集中式疏散交通信号优化模型，这种简化模型的求解效率较全优化模型在求解速度上提高83%；再根据优化后的交通流及动态可靠性指标反馈，设计过饱和交叉口群和致堵需求源的甄别治理技术，此中采用了自编码技术提高了甄别准确性；然后面向过饱和程度逐步降低的循环收敛要求，探索用于返回集中式控制的信号参数的固定策略，本课题后期采用的阀值控制策略使疏散时间较原始路网情形提高了7.6%；最终达成总疏散时间优化和过饱和预防双重目的。以多类突发事件和相异路网结构为例开展仿真测试与实证研究。