Traffic Status Mining And Prediction Method Based On Online Map Data

Online maps have become a part of our lives,a necessary tool for travelers to explore uncharted paths.However,outside of daily life,the existing research on online map data,especially the spatial features of traffic condition map,is not enough,and there is no self-contained traffic operation feature extraction method for online map data,which makes the online map data not fully and effectively utilized.There are hundreds of millions of users behind online maps.Systematic and reasonable data mining methods can further improve the theoretical level of urban traffic status identification and prediction.In order to solve the problem of online map data mining,this paper proposes an extraction method of traffic operation feature and its influence index data based on graphic data,which serves for the deeper traffic characteristic mining of road network,and tests the coupling of different traffic operation features with the method of multi-feature fusion prediction.This paper comprehensively explains the traffic function of online map from the perspective of travelers and managers,and analyzes the dependence and relationship between the data structure carried by online map and travelers’ demand.Based on the information guidance function of online map in social life,this paper makes a systematic introduction to relevant online map data.On the basis of obtaining online map data,the spatial and temporal distribution of traffic condition and traffic speed data is analyzed.Combined with genetic algorithm and the assumption that the short-term population in a large area is constant,the color weight of the population density in the heat map is calibrated and the corresponding weight results are obtained.In order to excavate the traffic operation characteristics in the online map data,this paper abstractions the spatio-temporal and location-related traffic slow queue attributes from the floating car tracks in the traffic condition map.The attribute of slow queue depends on a complete regional road network.Is relatively simple and easy access to regional road network,this paper path points to and geometric characteristics of online maps,object segmentation model and road pixels classification model is put forward,after the road object access road granular morphology is introduced,the sections of each segmentation object pixel refining until the extracted contour line of a single pixel width,characterized by sections of linear features.According to the starting point position bias set in the section classification model,the refined section contour points are reordered.By using the features of short links and ramps,the end points clustering by distance are successfully associated into several nodes,which are connected according to the corresponding number,so as to obtain a complete topological road network.On the basis of the complete road network,this paper successfully extracts the traffic operation feature of slow queue attribute from the online map data.The numerical example shows that the attributes of the main road slow queue can be obtained effectively.Combined with the two traffic operation characteristics and the influencing factors of population activities analyzed in this paper,the model structure of LSTM is optimized based on the input-output structure of the path,and a CNN-stacked LSTM model is proposed.The numerical example proves that the prediction accuracy of the neural network model is good,and it can be applied to the prediction of traffic operation characteristics.