Research On Stop Position Recognition And Travel Mode Analysis Based On Base Station Network Data

In recent years,with the rapid development of China’s social economy and urbanization process,the ownership rate of private cars is increasing,and people’s travel demand is also increasing rapidly,which leads to the increasingly serious problem of road congestion,and the traffic planning and management is becoming more and more important.Travel information collection is the basis of traffic planning and management.The traditional methods of artificial survey and equipment collection have small coverage and high cost.With the rise of new big data such as base station network data,it provides a technical means for accurately analyzing the relationship between residents’ travel behavior and the relationship between travel behavior and environment.For this reason,this paper puts forward the research of residence location recognition and travel mode analysis based on the base station network data.First of all,based on the summary of the development and architecture of mobile communication network and the common positioning technology of base station network,according to the generation conditions of base station network data,it can be divided into two types: event driven and network driven.The generation conditions of each type of data and the information contained in the fields are discussed,and the temporal and spatial characteristics of the base station network data of a City Unicom Company used in this paper are analyzed.Secondly,the noise data in the experimental data,including missing,redundant data,pingpong data,drift data,and its representation in the data set are studied in detail,and the corresponding threshold filtering rules are formulated to complete the data denoising.On the basis of data de-noising,this paper proposes a multi-mode combined method of residence position recognition.Firstly,each track point is identified according to the speed,then the user’s data track is abstracted as a weighted undirected graph with self loop,the user’s state is updated in the first round by using the community partition result of FN algorithm,and finally the state of track point is updated in the last round by using KNN algorithm.Compared with GPS data,92.33% of the distance between the start and end points of travel trajectory is within the service range of urban base station.Thirdly,based on the unsupervised fast k-Mediods algorithm and the supervised random forest algorithm,the travel mode analysis model is established.Through the analysis of the characteristics of base station network data and travel mobility characteristics,the appropriate feature parameters are extracted as the input of the analysis model.The parameters of the random forest model were optimized through experiments,and the two models were evaluated from the accuracy,accuracy,recall and F1 scores.It was found that the two models achieved high accuracy in the classification of three travel modes.Finally,according to the research of this paper,two kinds of traffic applications based on base station network data are proposed.The calculation method of travel sharing rate and the national traffic generation standard of health code are designed.