Research On Avoiding Spill-back Of Urban Traffic Congestion Control

With the development of society and the increase of urban population,traffic congestion has increasingly become the key issue in urban management today.Especially for some large and medium sized cities such as Harbin,Xi'an,Beijing,etc.,most of the sections are oversaturated during peak hours.The cause of poor road network performance(poor travel)may be an unreasonable signal strategy,but it may also be due to a traffic jam that the vehicle will not be able to travel normally(spillback in traffic),even if the signal light is green.How to control signal lights at intersections and avoid the appearance of traffic jam has become an urgent problem to be solved by researchers.From the macro-level requirements of urban traffic management,while reducing the waiting time for each vehicle is crucial,it is also important to avoid or reduce the occurrence of spillbacks across the entire transport network,especially on the main road.Therefore,for trafficsaturated cities,from the traffic manager's point of view,we should focus on avoiding the occurrence of spillback and pay more attention to the sections where congestion will or will have occurred.Therefore,while using the micro-comprehensive performance index of the entire network as an objective function of the optimization problem,the direct constraint on the queue length should be taken as the constraint of the road network optimization calculation.In addition,due to the increasingly sophisticated traffic monitoring equipment in recent years,the detection and storage of traffic flow and other data are becoming more and more common.More and more traffic-related studies are based on data.This article will also discuss how to use traffic data to establish a control model.For the problem that the computational complexity of the optimization process increases with the nonlinear model,we use a hierarchical structure to decompose the control problem and reduce the computational cost.In this paper,we propose a new urban congestion prediction control algorithm,the main concern is the saturation and over-saturation.The purpose of this study is to minimize the number of congested links in the network and to avoid or reduce the occurrence of spillback throughout the network by adjusting the length of the green light at each intersection.The main work of this paper is as follows:1)Propose a congestion control algorithm to prevent spillbacks.First,we emphasize hard constraints on queue length.Aiming at the problem of spillback,a hard constraint on the predictive value of queuing length is added to the optimization problem of predictive control,which directly avoids the occurrence of spillbacks.For situations that can not be avoided completely,that is,if there are too many constraints in the optimization problem and there is no feasible region,we need to set different priorities for different segments and relax the constraints according to priorities to obtain feasible domains and feasible solutions.SecondOptimize the road network equilibrium.Equalization terms are added to the objective function of the predictive control optimization problem.The objective function is to minimize the difference between the proportion of vehicles queuing in each section and the overall average of the road network.Control signals to balance the distribution of road network vehicles can fundamentally ease the queue queuing caused by the number of vehicles over the line back to the problem,improve road network utilization.2)Establish a neural network prediction model and use it for predictive control.Using existing data,a BP neural network is trained to predict the status of all road sections of the future road network from the current status and future control signals.In the process of optimization,the neural network is used to solve the optimization problem by back propagation.After the prediction is completed,the input end is continuously iterated and the gradient is decreased to find the feasible solution and the local optimal solution.3)Use hierarchical control structure to achieve congestion control of large-scale urban road network.A two-level control structure has been set up that takes the overall distribution of the road network as an overall goal.The upper controller optimizes the number of vehicles exchanged between sub-networks and provides guidance on the optimal number of queued vehicles on the sub-network.The sub-network optimizes the total number of vehicles queued inside the sub-network and makes the sub-network variables close to the upper set points.