| Intelligent connected transportation system is one of the most important development directions of intelligent transportation systems in the future.Intelligent connected transportation covers research areas,such as internet of vehicles and vehicle-to-road communications,autonomous driving,intelligent connected vehicles(ICVs),and automated highways.These techniques all have broad market prospects and important strategic significance.According to the communication protocols and data interaction standards,the communication and information sharing via vehicle-to-vehicle(V2V),vehicle-to-infrastructure(V2I),vehicle-to-road,vehicle-to-people,and vehicle-to-traffic management facilities or systems,are achieved to form intelligent traffic management and intelligent dynamic information services.The integrated intelligent network of ICVs for the future autonomous driving is an important application of the Internet of Things technology in the field of transportation.As a novel enhanced version of the travel tool,ICV is one of the most effective ways to solve traffic problems,including safety,congestion,energy consumption and environmental pollution.The lane-changing control for ICVs is the key technology to guide a safer,more comfortable,more energy-efficient,and more environment-friendly transportation system.Therefore,it has attracted widespread attention from the academic and practitioner communities.However,the current theoretical research on lane-changing control and evaluation of ICVs is insufficient.A comprehensive theoretical study on lane-changing control has not been well established,which restricts the development of the novel technique.On a theoretical level,this study can help better understand how lane-changing control of ICVs affects on freeway traffic flow,how lane-changing control algorithms affect the effectiveness of strategies for improving freeway safety and efficiency.On a practical level,the findings of this study can help promote the construction of smart highways,improve the level of intelligent connected traffic management,and reduce the existing severe traffic problems on freeways.The study tried to explore and analyze the critical issues associated with the lane-changing control and evaluation for ICVs from five aspects,i.e.,mechanism analysis,strategy comparison,effect analysis,technology application and prospective prediction.This work was sponsored by a National Key Research and Development Program and two National Natural Science Foundation projects.The study was conducted based on the American NGSIM database and transportation economy data collected from 18 countries.The study first established an ICV lane-changing model and three lane-changing route control strategies,and then proposed a machine learning-based model for lane-changing behaviors near an off-ramp.Additionally,this study applied evolutionary learning theory to establish a trajectory feedback control framework for ICV lane-changing process.At last,a safety evaluation index and a comprehensive economic evaluation model for intelligent connected transportation system were also developed.More specifically,this study includes the following contents:First,the study proposes the lane-changing model and route control strategies for exiting off-ramps.According to the traffic characteristics of off-ramp and ICV control,the study analyzes the effects of environment perception system on information collecting and lane-changing decision process.Also,the study explores the mechanism impact of lane-changing behavior near an off-ramp on the bottleneck traffic flow.After analysis of lane-changing gap calculation,prediction,comparison and execution steps during lane-changings,a mandatory lane-changing model for ICVs exiting off-ramps is developed.Various hierarchical lane-changing route control strategies are proposed,and the traffic scenarios under different control strategies are investigated and compared.Optimal lane-changing control strategy for exiting off-ramps on the multi-lane freeway is found out through comparisons between different scenarios.Second,machine learning algorithms are used to model the lane-changing behaviors near an off-ramp.The complete lane-changing process is divided into two steps: lane-changing decision and lane-changing execution.The study proposes a lane-changing decision model based on a random forest,and uses multiple vehicles and multiple gaps in the target lane to build a decision tree.Also,the study takes the impact of lane-changing behaviors on leading and following vehicles into consideration,and establishes a lane-changing execution model via back-propagation neural networks.Based on the differences in speed and position characteristics of lane-changing trajectories,the effects of the new lane-changing models on the intelligent connected transportation system are analyzed under different scenarios of the multi-lane freeway near an off-ramp.Third,the study establishes a feedback control framework for lane-changing trajectory based on evolutionary learning.The essence analysis and comparative study of evolution theory and traffic control are carried out,and then an evolutionary-learning-based control framework is proposed to solve the quantity and quality of lane-changing trajectory data for ICV research in the real word.Taking the lane-changing models based on machine learning algorithms as the simulation platform for intelligent connected environment,the study uses a comprehensive cost function to select the trajectory database and form the parent training set.Then the data set is re-extracted from the new and old lane-changing trajectory integrated databases.The study analyzes the impact of key parameters,such as the length of the lane-changing control area and market penetration of ICVs on travel efficiency and traffic safety.In addition,the study explores the effects of number of evolutionary learning iteration and lane-changing trajectories on the accuracy of fitting and calculation efficiency.Fourth,a safety evaluation index of lane changing behavior based on field theory is proposed.A lane-changing behavior modeling method is developed based on the interaction between five vehicles,and a comprehensive magnetic field of an ICV is constructed for the dynamic field and the behavior field.The lane-changing behavior evaluation based on the superimposed field force is calculated and the force is compared with the traditional index time-to-collision TTC.Moreover,this study determines the risk levels of lane-changing behavior with the even percentages as the setting principle.Besides,the study evaluates the speed evolution,field force distribution,risk assessment and other characteristics of different lane-changing route control strategies to provide decision support for the optimal strategy.Finally,a comprehensive economic evaluation model for intelligent connected transportation system is proposed based on the investigation of economic influencing factors of intelligent connected transportation system at the off-ramp bottleneck.After comparison between intelligent connected transportation systems with the conventional environment,the study collects the transportation-related economic data of 18 typical countries and 5 regions in the world.Meanwhile,the study predicts the various cost parameters in the comprehensive economic evaluation model from the prospective of short,medium and long terms.To assess V2V/V2 I control strategies,the study analyzes the economic performances of intelligent connected transportation systems in various countries and regions,and compares different control strategies in different periods.The findings of this study provide a theoretical basis for the economic evaluation and decision-making on control strategy for intelligent connected transportation systems. |
PDF Full Text Request