Bicycle Trajectory Data Cleaning,Matching And Riding Quality Analysis And Evaluation

In the context of the development of green transportation in China,traffic management departments need to quickly analyze and evaluate the riding status of road networks in order to implement targeted rectification measures.Based on the shared bicycle actual riding data,this paper analyzes the bicycle riding trajectory and the riding condition of the road section,studies the cleaning,noise reduction and matching methods of the riding trajectory data,and uses the processed riding data to realize the analysis and evaluation of the road riding quality,which provides data basis for the improvement of the bicycle riding road environment.The main research contents of this paper are as follows :(1)Analysis of problems and formation of research ideas.This paper analyzes the riding behavior law reflected in the riding trajectory data and the abnormal phenomena in the data.Based on the actual situation and characteristics of the trajectory data and the characteristics of bicycle riding,this paper puts forward the idea of using the riding trajectory data to realize the analysis and evaluation of road network riding condition and the key problems to be solved,including trajectory data cleaning and noise reduction processing,trajectory and road network matching,road riding quality analysis and evaluation,forming the overall research route and framework.(2)Research on cleaning and noise reduction of abnormal trajectory data.The source and performance of trajectory data anomalies are analyzed.Aiming at different types of anomalies such as ‘ pseudo trajectory data ’,sampling time anomaly data and positioning coordinate anomaly data,the anomaly data identification method based on boxplot and DBSCAN clustering is studied,and the process of trajectory data cleaning and noise reduction is given,which provides a reliable data basis for later research.(3)Research on the matching between trajectory data and road network data.According to the rider trajectory data after cleaning and noise reduction,considering the relative relationship between single trajectory point and road section and the relationship between multiple trajectory points,the matching model of rider trajectory and road network based on Bayesian network is constructed,and the matching probability calculation method based on piecewise rolling search is proposed.Each rider trajectory is matched to the corresponding road section in the road network,so that the later section can be used as a unit to analyze and evaluate the riding quality.(4)Road riding quality analysis and evaluation.This paper analyzes the bicycle riding condition on the road section,sets the judgment basis of riding quality,constructs the evaluation index system,and puts forward the evaluation method of road riding quality based on KNN algorithm.The relative evaluation of the riding quality of the road section is carried out,and the road section with poor riding quality is identified,which provides reference for the improvement of riding environment.(5)Taking the road near Beijing Xidan subway station as an example,the research method of this paper is verified and analyzed.