Research On Digital Twin Scene Generation Method Based On Complex Traffic Flow Network

With the development of science and technology,and the progress of society,more and more attention has been paid to traffic problems.The research based on complex traffic flow network has become a hot research topic.For the network,the scene generated based on digital twinning,as a key technology,plays an increasingly important role in the study of traffic flow.Based on complex traffic flow network,this thesis studies the generation method of digital twin scene,and proposes a digital twin model for analyzing complex traffic flow network.The main work of this thesis includes the following points:1.The traffic flow network of motorway is studied.Firstly,the data source used for research is introduced,and the research data is preprocessed.Based on K-means,correlation coefficient,trend change and external environment,the traffic characteristics of motor lanes were analyzed,and several evaluation indexes were proposed to evaluate the effect of the traffic flow network model of motor lanes,such as MAE,RMSE and MAPE.2.The traffic flow network of non-mixed traffic lanes is studied.Firstly,the characteristics of bicycle flow are studied,and then the characteristics of motor vehicle and non-motor vehicle mixed flow are studied,and the delay model of non-motor vehicle traffic flow blocking in non-mixed lane is proposed.Finally,based on the above research,the total delay model is proposed,and the relevant extension of this model is carried out.3.The generation model of digital twin scene is studied.First,the initialized data is generated from the description information based on the model input.Then,a single agent traffic behavior method is generated by reinforcement learning network.Finally,a digital twin scene is generated,and according to the digital twin scene,the relevant evaluation system and methods are determined,and the model is tested.