Warning Hotspot Identification And Influencing Factor Analysis Based On The Driving Data Of The Beidou Internet Of Vehicles

With the increase of road freight volume and freight turnover in China,the issue of road freight safety is becoming more and more important.Coupled with the encouragement and support of national level for the Internet of vehicles and intelligent networked terminal equipment,how to apply the Internet of vehicles and the technology of intelligent network to improve the safety level of road freight has become a new direction of traffic safety research.Based on the vehicle trajectory,operating status and warning information collected by the terminal of the Internet of Vehicles,this paper studies the temporal and spatial distribution characteristics of dangerous driving behavior warnings such as fatigue driving warning,speeding warning and collision warning,identifies the hotspot sections that produce warnings and reveals the main influencing factors,which could provide theoretical support for improving the safety level of road freight.The main contents of the paper include:(1)The extraction and processing of the driving warning data.Using Python programming and GIS map matching technology,a total of 16,560 pieces of fatigue driving warnings,speeding warnings and collision warnings were extracted from the original data set.The characteristics of space-time distribution are analyzed and the influence of time and space factors on the frequency of warnings is studied by analysis of variance(ANOVA).The results show that the time factor has a greater influence on the fatigue driving.The warnings of fatigue driving generated from 18:00 to 24:00 is significantly more than that in other periods.The impact of space factors is that the G4Beijing-Hong Kong-Macau Expressway produces significantly more speeding warnings than other regions within the city of Baoding,while collision warnings generated in Beijing is more than other regions significantly.(2)Warning hotspot sections identification and hotspots co-occurrence analysis.Using Moran's I index to analyze the degree of aggregation of warnings distributed in space,the results show that the three types of early warning are not randomly distributed on all roads under different time periods,weather and vehicle types.The road sections with more warnings have a certain degree of aggregation.By using the method of local spatial autocorrelation,the hotspot sections can be effectively identified.By calculating the co-occurrence index of hotspots in different time periods,different weather and different vehicle types,the results show that the hotspots of speeding warnings have a strong correlation with the time period of 12:00-18:00.(3)Analysis of factors affecting driving warnings.The traditional linear regression model(OLS),the spatial lag model(SLM),the spatial error model(SEM)and the spatial Durbin model(SDM)were used to analyze the potential factors that affect the driving warnings.The results show that the best fitting model for speeding warnings is the SLM,and the best fitting model for fatigue driving warning and collision warning is the SDM.It shows that the frequency of speeding warnings(dependent variable)is spatially related to the frequency of surrounding road sections,but its influencing factors(independent variables)are mainly related to its own section,and are not significantly related to the influencing factors of surrounding road sections.However,both the dependent variable and the independent variable of fatigue driving warnings and collision warnings have spatial lag effects.For example,when a toll station is set on a road section,the adjacent road sections are also prone to collision warning.