Research And Application Of Traffic Congestion Control Based On State Prediction

With the acceleration of urbanization,transportation has become an important factor affecting people's quality of life.Due to the increasing number of motor vehicles,the increasingly serious traffic congestion brings a lot of inconvenience to travel.In order to deal with traffic congestion,and improve the overall operation efficiency of the traffic network,the study of traffic congestion control has important theoretical and practical significance.Aiming at the problem of intelligent traffic congestion control,a traffic state prediction method based on spatio-temporal feature fusion in combination with machine learning and deep learning is proposed.Based on traffic flow distribution,signal optimization control is carried out for the congested bottleneck sections of the road network.The main work and contributions are as follows:1.A traffic state prediction method based on spatio-temporal feature fusion is proposed.The method is combined with traffic operation characteristics,feature engineering is used to determine the factors that affect the future traffic state of traffic flow,convolution neural network(CNN)has been used to extract the spatio-temporal characteristics of traffic flow parameters,long short-term memory(LSTM)has been used to extract the characteristic of traffic congestion sequence,after the channel feature fusion processing of CNN and LSTM,traffic state prediction is realized by using support vector machine(SVM).Based on the PEMS traffic data set,experiments were carried out and compared with other algorithms to verify the superiority of the CNN/LSTM-SVM feature fusion model.2.An optimization method of traffic jam bottleneck section signal based on traffic flow distribution is proposed.On the basis of analyzing the characteristics of traffic congestion,modeling analysis is carried out on the bottleneck sections,through the safety queue factor to deal with the future possible potential road congestion bottlenecks,through the residual capacity of road to adjust capacity allocation,traffic green split and phase difference of the optimization is completed.Through the simulation of VISSIM and the comparison experiment with other traffic control schemes,the superiority of the signal optimization of the bottleneck road section in this dissertation is verified.3.The urban intelligent traffic congestion control subsystem is designed and implemented.Based on the actual demand of the traffic state prediction and traffic congestion control,the urban intelligent traffic congestion control subsystem was designed and implemented,mainly including system management,congestion query,congestion control and traffic database,the core functions of the system were tested.The research results show that for traffic prediction and optimal control problems,the CNN/LSTM-SVM feature fusion model and the congestion bottleneck section signal optimization method based on flow distribution,to a certain extent solved the bad effects of traffic congestion,which has important theoretical and practical value.