Distributed Coordinated Control For Urban Traffic Networks

With the rapid development of human civilization and the sharp increase in the transportation demand,urban traffic congestion has become a tricky problem in metropolitan cities in recent decades.To alleviate the traffic congestion,one of the effective ways is by fully utilizing the available infrastructure with the use of traffic signals,ramp metering,the variable speed limits,route guidance and the other control measures.Since the urban traffic network is a large-scale complex system consisting of intersections,roads,and vehicles,it is essential to coordinate the various control measures.Urban traffic network is a typical large-scale complex system with high dimensionality,strong uncertainty,multiple objectives,and dispersed information.In this paper,model predictive control with rolling-horizon optimization is used in the on-line optimization to deal with the uncertainty and the multi-objective requirements.In addition,due to the dispersion of sensors and actuators in the transportation network,a distributed control scheme is proposed to avoid the huge communicational burden and computational complexity with a centralized controller.The overall optimization problem of the network can be decomposed into a number of suboptimization problems in the distributed control digram,which facilitates the local control of each agent with limited information and reduces the computational complexity.Nevertheless,the control actions of each subnetwork affect the performance of their neighboring agents owing to the couplings between them,and further affect the performance of the entire network.Therefore,how to coordinate the control actions of subnetworks to improve performance of the network is essential in the distributed control of the urban traffic network.This paper has proposed distributed control schemes for road networks with arterials,general road networks,and mixed urban and freeway networks,respectively.The main contributions of this dissertation are given as follows:· Propose a new measurement of the congestion degree of roads,i.e.,traffic process ability,which can be viewed as an indication of the utilization of the traffic resources and the coordination of traffic lights,and establish the mathematical model of the traffic light control system with the proposed measurement being the system state.Propose a "master-slave" coordinated control strategy for the traffic signal control in a network involving arterial roads,in which the interconnection of the traffic states and the control signals between the arterials and the side roads are considered.It can deal with the traffic signal optimization in a network with multiple cycle lengths of intersections and multiple arterials.As more comprehensive information is considered in the traffic process ability,the traffic signal control scheme presents better control performance.· Investigate the balanced control of traffic process ability for a general road network with a distributed control scheme.By achieving the local balance of traffic process ability of roads,the distribution of traffic is enhanced and the utilization of the traffic resources and the network efficiency is increased.Design a new method for the road weight,which can be used to recognize the congestion bottleneck based on the real-time traffic condition.The roads that have larger weights are served preferentially to accelerate the congestion dissipation.· Propose a distributed integrated model predictive control strategy and a two-level hierarchical control strategy for the mixed transportation network consisting of freeways and urban roads.In the traffic modeling of the mixed traffic network,a new method is designed to characterize the influence of the urban network congestion on the traffic capacity of the freeway network.Based on the traffic model involving the interconnections between the urban and the freeway part,a distributed coordinated control strategy is proposed to restrict the influence of the urban traffic congestion and the spillover of the off-ramps on the freeway traffic states,while reducing the computational and communicational burden of the centralized control strategy and making it easily implemented in on-line control.To enhance the control performance of the distributed control scheme,a perimeter controller is designed as an upper level controller to coordinate the transfer flow between the subnetworks and optimize the traffic distribution in the network.· Propose a distributed event-triggered control scheme for the mixed transportation network.To avoid the unnecessary updates of control signals,the event-triggered control mechanisms are applied in the traffic network.The triggered conditions are respectively investigated for the urban network and the freeway network due to their particular characteristics.Each agent determines the triggering of the updating of its traffic signals with the real-time traffic condition which eliminates the shortcomings of the time-driven control strategy.In addition,the triggering conditions of each agent can be designed according to the characteristics of the agent,which makes the control strategy more flexible and efficient.