Research On Detection Method Of Urban GNSS Anomaly Scene Based On Multi-source Spatio-temporal Data

The Global Navigation Satellite System(GNSS)has been widely used in military and civilian applications.GNSS navigation and positioning technology is the main means of high-precision outdoor positioning.The performance of GNSS directly affects navigation performance.In open scenes,GNSS signals are unobstructed and the quality of observation is good.However,when GNSS positioning is applied independently to tunnels,overpasses,tall buildings or dense building areas,areas with more vegetation coverage,lakes,etc.,GNSS signals are affected.High buildings,terrain,jungle occlusion or multiple reflections make the positioning accuracy worse or unable to locate.By excavating the GNSS anomaly area,a large-scale test environment is established for the next-generation Beidou chip test,which provides data support for the construction and improvement of urban road infrastructure,and provides a theoretical basis for the realization of accurate sensing and intelligent transportation for unmanned driving.The traditional detection method lacks a suitable method for obtaining a large number of signal anomalies around the road,and there are few existing methods for studying signal anomalies in a real environment.The results obtained by simply using the road surrounding environment data are difficult to verify.This paper studies the multi-source data such as floating vehicle trajectory data,road network data,geographic national survey data,building contour data,etc.,and designs an extraction algorithm for locating abnormal data,extracts the location anomaly points,and uses ST-DBSCAN space-time density clustering.The class clusters the abnormal points of the location,excavates the location of the anomaly,and combines it with the features of the surrounding buildings such as buildings,vegetation,water bodies,etc.,and excavates the GNSS anomaly area in the main urban area of Wuhan.The research work of this paper is as follows:(1)GNSS anomaly region extraction based on feature distributionWhen GNSS positioning is applied to challenging environments such as dense urban areas such as tunnels,overpasses,tall buildings,areas with large vegetation coverage,lakes,etc.,GNSS signals are occluded,multi-refractive,and reflected,resulting in effective data loss and signals.The intensity is weak.Dividing the elements of tunnels,overpasses,buildings,vegetation,water bodies,etc.from the national survey data,and combining them with the road network of the main urban area of Wuhan,dividing the main city into grids and surrounding the grid area and roads.The geographic element data and the building outline data are superimposed,and the information of the object elements such as the object area or the height of the building which is likely to cause GNSS anomalies in the statistical grid is statistical.The evaluation index is determined,the GNSS anomaly regional evaluation model is constructed,the index weight is determined by the entropy weight method,and the GNSS anomaly area in the main urban area is evaluated based on the evaluation model.The specific research contents include: 1)factor division based on geographic national census data;2)geophysical element statistics based on grid analysis;3)GNSS anomaly area extraction in main urban area based on evaluation model.(2)GNSS anomaly region extraction based on location anomaly trajectory point clusteringThe floating car trajectory data is information such as the position,time,and speed direction of the moving object at equal time intervals.In the complex environment of urban areas,the two main factors affecting GNSS performance are occlusion and multipath,which causes the receiver to not receive signals or signal drift.It is reflected in the positioning data that there is no data acquisition or drift at a certain positioning point..The sampling time interval and the normal moving distance threshold within the sampling time interval are determined.On this basis,an extraction algorithm for locating anomaly data is designed to extract the location anomaly.The floating vehicle trajectory data and related behavior trajectory data are spatio-temporal data that changes with time and have spatio-temporal correlation.Spatio-temporal clustering is an important method for mining spatiotemporal clustering patterns with strong spatiotemporal correlation.The ST-DBSCAN space-time density clustering algorithm is used to cluster the abnormal points and mine the abnormal regions.The specific research contents include: 1)time-constrained signal missing abnormal point extraction;2)distance-based signal drift abnormal point extraction;3)GNSS abnormal region extraction based on localized abnormal point clustering.(3)Mining and analysis of GNSS anomaly regions based on multi-source data fusionn this paper,the areas where GNSS anomalies are prone to occur are judged by combining the overpasses,tunnels,buildings around the urban roads,vegetation,and water bodies.The GNSS anomaly region obtained based on the feature distribution and the GNSS anomaly region obtained by clustering the anomaly trajectory points are comprehensively analyzed,and the GNSS anomaly region in the main urban area of Wuhan is obtained,and the signal anomaly region around the road is obtained in a wide range.