Research On Traffic Congestion Pattern Recognition Based On Taxi Data

With the continuous development of urbanization and motorization,The traffic congestion has become a challenge faced by various cities.Traffic congestion can be evaluated with multiple indicators,and the basis for calculating the indicators lies in data collection.Compared with traditional methods,using on-board GPS equipment to collect taxi data has many advantages.Based on the big data of taxi trajectory and road network data,this paper evaluates the traffic congestion in Hangzhou,and recognizes the congestion patterns to provide ideas for the management of traffic congestion in Hangzhou.The main research contents of this paper are as follows:(1)This paper designs a preprocessing method for taxi data.The research data in this article includes the GPS data of taxis from December 4,2017 to December 10,2017,along with the road network data in Hangzhou.Due to the quality problems in the taxi data,this paper designs a data cleaning process to eliminate quality problem interference,and use the road matching algorithm to link the taxi data with the road network data.(2)This paper designs road congestion evaluation method.According to the characteristics of the road network data,this paper makes a more refined classification of road sections.In terms of the relevant theories of road engineering,this paper calculates the frequency distribution of vehicle speed on each level of road section,and determines the classification standard of congestion score corresponding to each grade of road.According to the speed of the vehicles driving on each road segment,the whole day is divided into multiple time periods with a 15-minute statistical interval,and the average speed of each road segment in each time period is calculated.This paper calculates the congestion level of each road segment in each statistical interval through statistical average speed data,and uses the INRIX Index algorithm to compute the congestion index of individual roads and the overall road network in each statistical interval.(3)This paper analyzes the temporal and spatial distribution characteristics of traffic congestion in the study area.Based on the road congestion evaluation index,it analyzes the temporal and spatial characteristics of traffic congestion in Hangzhou.In terms of spatial distribution characteristics,the congestion in Hangzhou is concentrated in four regions,and its spatial distribution status is constantly changing over time,which has different characteristics during the morning and evening peaks of working days and the morning and evening peaks of holidays.In terms of time variation characteristics,the overall congestion index of Hangzhou has a "double peak" intraday pattern on weekdays,but only a "single peak" on holidays,occuring during the evening peak period.(4)Congestion pattern recognition of frequently congested road sections.In this paper,firstly,according to the length of time that the road section is in severe congestion in a single day,the frequently congested road sections are selected.Then,cluster analysis method is used to identify the traffic congestion pattern on frequentlyoccurring congestion roads.It constructs a cluster analysis model of traffic congestion patterns,including determining the clustering algorithm,clustering index and the method of obtaining the optimal number of clusters.Subsequently,cluster analysis was performed on the daily variation characteristics of the congestion index of each frequently congested road section,and the congestion mode of the frequently congested road section is divided into six categories according to the results.Based on the analysis of the characteristics of each mode,they are summarized as multi-peak type,low front and high back type,sharp type,high front and low back type.This article describes the characteristics of each mode of congestion,and analyzes the actual sample road sections,and provides a certain idea for the traffic management department to formulate different governance strategies for different types of congestion.