Research On Control Of Urban Traffic Network Congestion

Model traffic has made people live more conveniently.However,as urban traffic network expands,the increasing traffic demand can not be satisfied by infrastructure constrution like widening roads or building viaducts.Traffic jam has been a huge problem to solve.As a traditional control method,fix-timed control strategy is able to keep vehicles in order.But,the open-looped control theory has not considered traffic network's structure or its current state.It can not solve the traffic jam problem.Classic centralized control theory for large urban traffic network has been widely discussed.Based on nonlinear model predictive control,these kind of control method can predict future network state with macroscopic models and optimize current traffic condition of network using traffic signal.The disadvantage of centralized control method is its computational complexity.The computational time will rise exponentially as the traffic network grows,due to the complicated traffic model,which makes it difficult to be applied realistically.Recently,urban traffic control theory with a multi-level hierarchy has been proposed.It lowers computational complexity by dividing traffic network into parts,but still the computational efficiency is not enough for real-time control.Ideal traffic control method is supposed to ease traffic congestion while mantaining high computational efficiency at the same time.Here are several approachs to achieve it:· Simplifying the predictive model.Computational efficiency can be increased by lower the calculation of traffic model.· Adopting new control methods.Traditional model predictive control methods uses objective function and constraints to achieve optimization.By removing the objective function and concentrating on constraints of congestion,higher computational efficiency might be obtained.Based on previously described thoughts,the main work of this article is consisted of following parts:1)A simplified traffic network model is proposed based on the S model in [1].It improves computational efficiency while maintaining certain accuracy and can predict future network states given current traffic conditon.This model is the basis of traffic congestion control which will be discussed later.2)The traffic congestion control theory is proposed.Instead of optimized solution,feasible solution is computed with the help of simplified predictive model.Apparently,computational efficiency is increased when computing feasible solution,and it is more understandable when vehicle number in every road of traffic network is used to describe current congestion state.3)Simulation of control method with a multi-level hierarchy is run and the control effect is verified.This article discussed the effct of upper layer which balances traffic demand of different subnetworks and find out that a proper weight of upper layer can achieve similar control effect as centralized control method with much lower computational comlexity.