| With the continuous development of social economy and the rapid increase in vehicle ownership,traffic congestion on urban expressways has been widely concerned.With the development of navigation applications,it is possible to obtain massive new types of traffic data,and in recent years,intelligent analysis technology has been widely popularized.This study explores new ideas on the causes and prediction of traffic congestion on urban expressways based on navigation data by adding new traffic data provided by navigation systems to the original road infrastructure and traffic facilities,which provides theoretical reference and technical support to reduce road traffic congestion,accidents and pollution and improve the operational efficiency of expressways.It can also improve residents’ travel environment and road service quality and play an important role in helping urban traffic management.The thesis focuses on two typical sections of urban expressway diversion area and exit sections with frequent congestion.Based on navigation data and in combination with multi-source data such as traffic facilities and road conditions,the thesis conducts research on feature mining,cause analysis,and congestion prediction of traffic congestion.The main research work of the thesis includes:Firstly,in the typical road segment scenarios of expressway diversion areas and exit sections,a comprehensive analysis database of traffic congestion based on multisource data fusion is established and the evolution characteristics of traffic congestion under different conditions are analyzed.This research uses navigation to collect traffic operation status,time,road ID data,etc.,matches multi-source data such as transportation facilities and road conditions collected on the spot,and establishes a comprehensive analysis database of traffic congestion under the two scenarios,and this paper conducts research on the spatiotemporal distribution law of traffic congestion to obtain the evolution characteristics of traffic congestion under different space-time and weather conditions.Secondly,in order to explore the impact of multidimensional factors on urban expressway traffic congestion,as well as the correlation and co-occurrence laws between the two,a structural equation model and association rule algorithm are used to conduct a comprehensive causal study on traffic congestion.Research is conducted to construct a structural equation model to screen various influencing factors,and analyze their impact on traffic congestion.Selecting factors that have a significant impact as key causal indicators,a traffic congestion correlation model for urban expressways is established based on the targeted optimization of the Apriori association rule algorithm.The correlation and co-occurrence laws between multiple key causal indicators and road congestion status are analyzed,and a research method for the causes of traffic congestion on urban expressways is developed.Finally,in order to accurately predict the traffic congestion status of urban expressways,traffic congestion prediction models are established for typical road sections in the diversion area and exit section of urban expressways.The research analyzes the impact results of various causes,takes the key factors of traffic congestion as the key variables of traffic congestion index prediction,introduces the XGBoost model improved by sparrow search algorithm to establish the urban expressway congestion index prediction model,and compare with the PSO-XGBoost,SVR,random forest model and relevant literature modeling results.The results show that the improved XGBoost prediction model has higher prediction accuracy and applicability.In summary,this paper focuses on the causes and prediction of road congestion based on navigation data for typical congested road scenarios in expressway diversion areas and exit sections,providing new ideas and approaches for improving road congestion and providing theoretical reference for road traffic management. |
