Research And Implementation Of Urban Road Travel Time Prediction Model For Intelligent Transportation

In recent years,Intelligent Transportation Systems(ITS)have played an increasingly important role in traffic operation and management.Among them,predicting urban road travel time is of great significance for optimizing road planning and management.Due to the instability and uncertainty of traffic,traditional prediction methods based on predetermined probability distribution models cannot fully utilize the topological information and uncertain features in road networks,thus affecting the accuracy of the prediction results.To address this issue,this thesis proposes a Bayesian spatiotemporal graph convolution-based urban road travel time distribution prediction model.The main research contents of this thesis are as follows:(1)To conveniently obtain a large amount of real vehicle travel data in road networks,this thesis first proposes an interactive voting-based map matching method,MIVMM,which matches GPS trajectory points to urban road networks and obtains a large amount of historical travel speed data for city roads.The map matching method adopts 3D kd-tree,Haversine formula,A-star shortest path algorithm,and Web Mercator projection techniques to analyze candidate points in both time and space,achieving efficient and accurate map matching.(2)Based on the obtained matched data,this thesis proposes a Bayesian spatiotemporal graph convolution-based urban road travel time prediction model.This model fully integrates deep learning,graph convolutional neural networks,Bayesian theory,autoencoders,and generative adversarial networks,effectively estimating the distribution of urban road travel time.Meanwhile,the model demonstrates good scalability and robustness when dealing with a large amount of urban road data.(3)By implementing the Bayesian deep graph convolution-based urban road travel time prediction model,a new solution is provided for the distribution of urban road travel time.This study not only helps improve the efficiency of road planning and management but also provides significant support for the development of intelligent transportation systems.Moreover,incorporating other traffic data sources(such as traffic signal control and traffic monitoring cameras)can further enhance prediction performance,offering more comprehensive support for optimizing intelligent transportation systems.