Traffic Feature Analysis And Traffic Flow Forecasting Based On Spatio-temporal Data

In recent years,the process of urbanization has continued to advance,the scale of cities has continued to expand,and traffic congestion has become more and more serious.The large-scale application of spatio-temporal data and the emergence of smart transportation platforms provide new ideas for alleviating urban traffic congestion.The traffic spatio-temporal data is rich in content,current and highly accurate.The application of spatio-temporal data in the transportation field can relieve traffic congestion and provide technical support for the traffic management department.Based on spatio-temporal data,this paper explores the temporal and spatial characteristics of traffic and congestion patterns in the main urban area of Xi’an from a temporal and spatial perspective,uses deep learning methods to predict traffic conditions on Taihua North Road,and introduces traffic simulation technology to determine the frequent occurrence of the study area Congestion mitigation strategies are proposed.The main research contents are as follows:(1)This paper studies the temporal and spatial characteristics of traffic congestion in Xi’an,and finds that traffic congestion mainly occurs around weekdays and holidays.Within a week,the congestion on working days is more serious than on rest days,and the traffic congestion index on working days is 11%higher than that on rest days;due to the impact of the new crown epidemic,compared with historical data for the same period,traffic congestion in the main urban area of Xi’an in February 2020 The index has fallen sharply.After March 2020,the traffic index has rebounded significantly,basically the same as the historical data for the same period,indicating that the traffic flow affected by the epidemic has basically returned to a normal level.The thesis integrates the characteristics of traffic flow in time and space dimensions,and proposes a traffic jam mitigation strategy that improves the functions of each area and realizes the balance of work and housing in functional areas.(2)According to the traffic conditions of Taihua North Road,this paper establishes a deep learning model.The parameters that need to be set in the model include:the number of neurons in the input layer,the number of neurons in the output layer,the number of neurons in the hidden layer,and the number of network layers.After the modeling is completed,explore the influence of various parameters on the prediction accuracy of the model.After continuous experiments,this article has obtained the optimal structure of the deep learning model.The specific structure is 264-82-30,that is,the number of neurons in the input layer is 264,hidden The number of neurons in the layer is 82,and the number of neurons in the output layer is 30.After calculation and verification,the prediction accuracy of the model is 92.4%.(3)Based on the prediction results,the strategy proposed in this paper for the traffic unit of Taihua North Road is as follows:by adjusting the operation mode of traffic lights,the traffic flow is diverted,and the lane occupancy rate is used as an evaluation standard.At intersections,this strategy can reduce the average lane occupancy rate of the road from 40.6%to 31.3%,while the average lane occupancy rate of the total road network is reduced from 36.4%to 34.1%.It can be seen that the program has a certain blocking effect.