| With the rapid economic growth,our country’s highway construction has also developed rapidly.At the same time,highway congestion is becoming more and more common,and traffic congestion will cause a series of problems such as reduced traffic efficiency and environmental pollution.Therefore,it is of great significance to study the specific influence of influencing factors on expressway traffic flow congestion and analyze the causes of traffic flow congestion,so as to improve the efficiency of expressway management and control and alleviate traffic congestion.This paper starts with a qualitative and quantitative analysis of highway traffic flow congestion,acquires and processes highway multi-source data,and identifies traffic flow congestion data to construct a traffic flow congestion information file.On this basis,the highway traffic Analyze the causes of congestion.The research work of this paper mainly includes the following aspects:(1)Multi-source data collection and processing.Through qualitative and quantitative analysis of traffic flow congestion on expressways,the influencing factors of traffic flow congestion are discussed from three aspects: road conditions,traffic flow and time factors;Based on the road segmentation of the highway combined with important road structures(such as tunnels),the average travel time data of each road segment is obtained based on the Gaode Map developer platform path planning API,and the traffic status of the road segment itself and adjacent road segments is analyzed.On the premise of establishing outlier identification and processing methods;Calculate the crosssectional flow by using network charging data;Use Auto CAD software to extract turning radius.(2)Identification of traffic congestion.By establishing the identification method of high-frequency congested sections of expressways,the spatial and temporal characteristics of expressway congestion are analyzed;According to the change characteristics of road traffic and average speed at the time of accident,establish a traffic flow congestion discrimination method based on the abrupt change of speed,use statistical analysis method to determine the threshold range of the index,and verify the analysis,the identification method can better distinguish the traffic type and event-based congestion;On this basis,the real variables and virtual variables are introduced to construct traffic flow congestion information archives.(3)Analysis of the causes of traffic flow congestion.Based on SHAP framework and regression analysis method,the cause of traffic flow congestion is analyzed.Use the XGBoost algorithm to sort the importance of the factors,and interpret the results of XGBoost in combination with the visual analysis of the SHAP value;Through the regression analysis method,find the combination of factors with the highest degree of interpretation,and construct a regression model of the combination of important factors for analysis.The analysis results based on the SHAP framework and regression analysis method show that:Factors such as flow,length of road section,proportion of large vehicles and time period are important factors that affect traffic flow congestion;The former quantitatively analyzes the threshold of each factor from the overall road level,while the latter can analyze the impact of multiple factors on traffic flow congestion,and quantitatively analyzes the threshold of each factor from a more microscopic road level;For the overall data set,the XGBoost model has a better estimation effect,and the important factor combination regression model has a better estimation effect when there is congestion.Both models have good reliability. |
