Modeling And Control Approach On Expressway Based On Macroscopic Continuous Flow Model

With the rapid increase of motorized traffic demand during the past decades,urban expressways have been suffering from unpredictable traffic congestion and safety problems.Macroscopic traffic flow modeling help describe and reproduce traffic flow dynamics of freeway networks.It is also beneficial for understanding the mechanism that traffic congestion is created and propagates.In addition,efficient traffic control can relieve the pressure of traffic congestion and improve traffic safety conditions.However,existing macroscopic traffic flow models were developed and operate under the assumption that the modelled traffic networks are homogenous in traffic flow dynamics with complete boundary information available.Moreover,traffic control strategies that are evaluated to be fine based on macroscopic traffic models may often perform worse in practice.In view of all these,this dissertation aims to explore a few key issues concerning macroscopic traffic flow modelling and control technologies for freeway traffic,in four aspects:mechanism,technologies,performance and practical examplesRegarding traffic flow modeling,model calibration and validation was firstly studied with respect to a large scale inhomogeneous expressway network in Shanghai.The corresponding scientific problem is defined as an optimization problem for a nonlinear dynamic system,and a variety of nonlinear optimization algorithms were employed to the solution of the problem.The sensitivity of the calibrated model was evaluated with respect to the complex topology and different weather conditions involved.Regarding the flow estimation for ramp pairs without installing detectors,the dissertation proposed a mathematical model describing the connection between ramp flow and mainstream traffic state,by which machine learning and deep learning methods were applied to identify intrinsic relationship between missing ramp flows are mainstream traffic state.Then missing ramp flow estimation problem was solved with satisfying performance.Regarding the control problem,the dissertation conducts qualitative shockwave analysis and quantitative evaluation based on a representative case study models at beginning.Then discuss the significance of implementing traffic control through the case of ramp metering.After that,the dissertation addresses a local ramp-metering problem in the presence of far-downstream bottlenecks,and explores feasibility of PI-ALINEA with exhausting simulation experiment and field application results.Among them,the simulation research constructed three distinct cases of bottleneck that may often be encountered in practice,and the field test scenario is in the Monash highway in Melbourne.Finally,to improve the expressway efficiency,the dissertation proposes a novel variable speed limit strategy for maximizing bottleneck throughout.After simulation verification,the strategy was applied to the SR 78E freeway in San Diego,California.Then a new assessment method of driver compliance rate was constructed to quantitatively evaluate the control strategy performance based on the practice data.From the perspective of theoretical research,this study focused on the key problem of network traffic flow modeling,which helps to reproducing traffic phenomena and understanding the mechanism of complex traffic characteristic evolution.Research on ramp flow imputation problem makes it effective to solve the longstanding problem of system unobservability.From the perspective of engineering application,this study pioneering utilized city freeway data on national scale to complete traffic dynamic modeling,calibration and validation work.The ramp metering algorithm and variable speed limit strategy have successfully applied in freeways across Australia and United States.The results provide theoretical and experiential basis of subsequent traffic management research,which is beneficial for improving expressway traffic management level and solving the severe traffic problems in existing expressway network.