Research On Road Traffic Flow Prediction Model And Road Network Flow Optimization Based On Deep Learning

In recent years,with the rapid development of economy and society,the number of motor vehicles has also increased dramatically,and urban traffic congestion has become increasingly serious.In order to effectively support residents 'travel choices,urban traffic management departments' traffic control and trip guidance,short-term traffic flow predictions have become countries Hot issues for scholars.Short-term traffic flow has the characteristics of real-time,non-linearity,periodicity,etc.It is a typical time series data.Obtaining the potential correlation and influence between data is the key to accurate prediction.Deep learning can obtain the characteristics of the data well by using a multilayer network structure.Based on this,this paper studies the short-term traffic flow prediction method of deep learning.The main research contents are as follows:First,the improved particle swarm optimization least squares support vector machine algorithm(IPSO-LSSVM)is used to predict the future path traffic flow,chaos is used to initialize the population,and the variable weight combination algorithm is used to optimize the local and global search capabilities of the particles.The time series of Qianlingshan Road,Yunyan District,Guiyang City,Guizhou Province was taken as the research object,and comparative verification and simulation analysis were performed.Because the traffic prediction of a single path cannot solve the traffic congestion problem,based on this,the traffic prediction and traffic regulation of the traffic network are particularly important.Secondly,in order to further improve the accuracy of road traffic network traffic flow prediction and make full use of the spatial correlation of road traffic flow,a prediction model for generating adversarial networks was proposed.The idea of non-cooperative game-"maximum and minimum" equilibrium was used to make predictions.The data processing results of the model approach the distribution of real traffic flow data,thereby improving the accuracy of traffic flow prediction.Finally,traffic prediction is the means,and road network traffic optimization and balance is the key.Based on the previous prediction results,using the principle of price leverage in economics,the road network path charging method is used to achieve the optimal distribution of road traffic.Further,in order to distinguish planned trips from unplanned trips and maximize social benefits,only the non-bus portion of the road traffic flow is charged,and a mixed integer linearity is established with the minimum overall travel cost of road network travelers as the objective function The planning model is analyzed with numerical examples.Simulation results show that theproposed charging strategy can achieve the optimal distribution of traffic flow and reduce congestion.