The Research For Dynamic Multi-objective Optimization Control Of Expressway Considering Emissions And Fuel Consumption

As people's increasing demand for driving,the expressways have rapidly reached saturation and traffic jams often occur.In China,reoccured congestion takes up a large proportion,causing a waste of time and economic loss.In addition,due to serious environmental pollution and resources shortage,it is significant to pay much attention to emissions and fuel wastage resulted from traffic congestion.Therefore,it is important to handle congestion through reasonable traffic control.This paper proposes a more scientific and effective control method to improve expressway's efficiency,and reduce pollution and energy wastage.This article mainly talks about expressway traffic control.It uses macroscopic traffic flow models to simulate traffic behavior,and an integrated control strategy with variable speed limit and ramp metering for expressway control.In terms of traffic control methods,this paper explores a general framework including dynamic multiobjective optimization algorithms and model predictive control theory.The main work and achievements of this paper are summarized as follows:1)This paper first summarizes relevant researches on expressway traffic control,and reviews the common control methods.On account of a good robustness and control performance,this paper mainly focuses on model predictive control.However,to some extent,this method is not effective to sovle expressway traffic control such dynamic and real-time optimization control problem with multi-objectives.Therefore,it reviews the acheivements of dynamic multi-objective optimization algorithms,hoping to adopt a dynamic multi-objective optimization algorithm to handle traffic optimization control.2)This article introduces the basic concepts of model predictive control and its application framework in traffic control.Then it simply explains the control strategy adopted in this paper and several common-used performance indicators,which are also considered as optimization objectives in model predictive control.Based on model predictive control,this paper abstracts expressway traffic control into a dynamic multiobjective optimization problem,and then illustrates the concept of dynamic multiobjective optimization problem.Thus,this paper adopts a dynamic multi-objective optimization algorithm to solve expressway traffic control,and designs a new model predictive control framework based on dynamic multi-objective optimization algorithm.3)Based on above framework,this paper explores two dynamic multi-objective optimization algorithms.First dynamic optimization algorithm is based on clustering prediction,which is extended from NSGA-II.The algorithm performs good both in convergence and distribution in three different test functions.In order to verify its performance in traffic control,this paper executes a simulation study on the basis of above framework through a single-ramp expressway network.This algorithm is used to optimize traffic control solutions.The traffic flow model is METANET,and a method called TOPSIS is applied to deciding optimal control solution.Simulation results show that the model predictive control based on this algorithm can effectively alleviate traffic congestion,emissions and fuel consumption.4)The second dynamic multi-objective optimization algorithm is based on multipopulation prediction,which combines multi-population and prediction strategies.It is other extension of NSGA-II.Its performance evaluation shows a good convergence,so it is also applied to solve traffic optimization control under proposed framework.This paper selects a real expressway system for further simulation study.It is a section of Guangzhong Road in Shanghai.Here,CTM model is used for simulating traffic behaviors.Due to more than one ramp in the expressway,it leads to an increase in calculation time.Thus,this paper adopts a distributed control structure,then makes a comparison with integrated control structure.Besides,this algorithm is also compared with the first algorithm.The simulation results show that the model predictive control methods based on two optimization algorithms have a better control performance.The performances are close between two control methods with the same control structure,while there is a difference between different control structures in each method.5)In order to achieve on-line optimization controls in above simulation studies,this article designed a platform.This platform is mainly used for data transmission between two different tools,which are applied for optimization and control respectively.It can also save relevant data into database for further data analysis.