Research On Fast Search Clustering Algorithm Based On Taxi Trajectory

With the rapid development of satellite positioning technology and wireless communication technology,more and more mobile positioning devices are installed on moving objects,and the acquisition of GPS trajectory data has become easier and easier.As a result,a large amount of trajectory data has been generated.How to mine valuable information from a large amount of trajectory data has become very important.At present,the more popular trajectory data mining direction is trajectory clustering.Through trajectory clustering,the movement rules and behavior characteristics of these moving objects are found,which provides some references for business decision-making and urban development planning.As an important part of urban transportation,taxis have the characteristics of wide distribution,large amount of data,and easy access.At the same time,taxis are the main choice for people's transportation,and their data can well reflect people's travel laws.This article takes Beijing taxi data in 2012 as an example.Through theoretical research on trajectory data processing,trajectory similarity measurement,and trajectory clustering methods,the data is filtered,reduced in noise,and compressed,and then processed.The trajectory similarity calculation method combines the characteristics of taxi trajectories,selects an appropriate distance measurement scheme,calculates the similarity distance between trajectories,and finally clusters the trajectory data according to the similar distance to analyze the hot spots of the city.The specific research work of this paper is as follows:(1)High-precision compression algorithm based on taxi trajectoryIn this paper,in the processing of trajectory data,in addition to reducing the longitude of the trajectory data,filtering the trajectory data,and sub-trajectory division,the Douglas-Peucker trajectory compression algorithm is improved.Coordinate transformation is added to the trajectory compression algorithm.To make it possible to compress GPS track data.In the process of trajectory compression,identification and deletion operations of noise data are added.Experiments show that the improved compression algorithm,while further improving the compression efficiency,retains the shape of the trajectory well and ensures the accuracy of the data.(2)Fast search clustering algorithm based on taxi trajectoryAiming at the problem of relatively unreasonable selection of truncation distance in the fast search clustering algorithm based on density peaks,this paper refers to the method of determining parameters in the data domain and combines the characteristics of taxi track data to improve the fast search clustering algorithm based on density A more reasonable truncation distance is determined by means of potential energy entropy.The experiment proves that this method can well select the clustering center.(3)Analysis of Urban Hotspot PathsThis paper compares the clustering results of different trajectory similarity measures and finds that the trajectory similarity measure scheme SSPD is more suitable for this paper.By clustering trajectory data in different time periods,analyzing urban hotspot paths,digging people's travel rules,and discovering urban hotspot passenger carrying paths,it is convenient for relevant departments to allocate transportation resources more reasonably.