Spatial-temporal Distribution Characteristics And Prediction Of Urban Road Average Travel Speed Based On Traffic Status Data

With the rapid advancement of urbanization in our country and the large number of local government policies to attract talents,the number of urban population and cars have increased year by year,while the growth of urban road construction mileage has lagged far behind,resulting in a growing contradiction between traffic supply and demand,the Traffic congestion is intensifying,and the direct performance of traffic congestion is that the travel speed of different grades of roads is too low,so the average travel speed of roads is closely related to traffic jam.Therefore,the spatial and temporal distribution characteristics and influencing factors of average driving speed are explored from the city level.It is necessary to make predictions.It is difficult to obtain the travel speed data of the entire urban roads by using the traditional measurement method.With the continuous accumulation of technology,the development of technology companies such as smart phone navigation and the Internet is changing with each passing day,resulting in a large amount of valuable data and more advanced data collection methods.The means of collection made it possible to study the average travel speed of the road from the city level.This paper firstly uses the Python language to call the Amap API interface through programming,so as to obtain the average travel speed of 254 roads in Xi'an City Expressway every 5 minutes,a total of 4 weeks of average travel speed data;then use cluster analysis to collect Divided into three categories of normal working days,rest days and holidays,visualized by Arcgis10.5,and with correlation coefficient,Fourier Fast Transform(FFT),similarity coefficient,trend analysis,Moran's index,hotspot analysis,etc.A variety of methods were used to analyze the distribution characteristics and data characteristics of road average travel speed under different categories of time and space.Then,combined with various auxiliary data such as distribution of business circle,different limit conditions,population density of each street,road network density,bad weather in rain and snow,and before and after the opening of the subway,the influence degree of different external factors on the travel speed of the road segment is analyzed and significant.At the same time,from the time and space dimensions,the time state vector of the K-nearest neighbor prediction model is improved to consider the space-time state matrix of time and space,and the measures such as exponential weighted time weight and Gaussian weighted space weight are proposed.The prediction model is improved.The results show that the average travel speed data has similarity,periodicity,space-time correlation and spatial unsteadiness.The average travel speed is lowest during the late peak period,and the road segment with higher TTI index is concentrated on the expressway and trunk.Roads,traffic transition nodes and areas with high population concentration,and show obvious moving characteristics during the morning and evening peak hours;business circle distribution,different limit conditions,street population density,road network density,rain and snow bad weather on the road The average travel speed has different degrees of influence.The subway opening has no significant effect on the average travel speed of the roads along the road.By comparing the models before and after the improvement,the improved K-nearest neighbor prediction model is simple and effective.