Analysis Of Motorcycle Traffic Accident Injury Severity Considering Spatial And Temporal Heterogeneity

One of the important research directions of road traffic safety is the analysis of influence factors of traffic accident severity.Motorcycle is a weak party in traffic participation.When motorcycle traffic accident occurs,the driver will be directly hurt.Most of the existing studies only consider the influencing factors of the severity of motorcycle traffic accidents,while ignoring the temporal and spatial information of traffic accidents.However,time and space have an important impact on the severity of traffic accidents.Therefore,from the perspective of spatial and temporal heterogeneity,this study explores the regularity of influencing factors of motorcycle traffic accident severity in space and time,in order to improve the level of road traffic safety.Firstly,in order to solve the problem of spatial heterogeneity scale of motorcycle traffic accidents.Based on the Geographically Weighted Regression model,a Multi-scale Geographically Weighted Regression model is constructed by relaxing the assumption that the spatial process scales of all local coefficients are the same.The spatial scale of different influence factors is studied by MGWR model.Combined with the spatial distribution map of regression coefficient,the spatial heterogeneity of influence factors on the severity of motorcycle traffic accidents is analyzed.Secondly,in order to solve the problem of temporal and spatial heterogeneity of motorcycle traffic accidents.By increasing the time information when traffic accidents occur,the Geographically and Temporally Weighted Regression model is constructed by using the combination of double square weight function or Gaussian function,adaptive bandwidth or fixed bandwidth respectively.The optimal combination of space-time weight function and bandwidth is explored,and the model results are tested for space-time nonstationarity.Based on the spatial and temporal distribution map of regression coefficient,the influence of factors on the spatial and temporal heterogeneity of motorcycle traffic accident severity is analyzed.Finally,for the widespread group heterogeneity in motorcycle traffic accidents.The accident data are divided into two categories : young drivers and elderly drivers.The correlation between age and the severity of motorcycle traffic accidents is explored.Multi-scale Geographically Weighted Regression model and Geographically and Temporally Weighted Regression model are used to construct traffic accident severity models for young drivers and elderly drivers respectively.The spatial and temporal heterogeneity factors identified by the model are compared and analyzed.The results show that:(1)Multi-scale geographically weighted regression model can effectively identify the spatial scale of influence factors.According to the regression results of the model,8 influence factors’ spatial scale are small,indicating that the spatial heterogeneity are large.Including fatigue,strong light,speed,number of vehicles involved,town,T-intersection,no light at night and cloudy.(2)The optimal geographically and temporally weighted regression model can be constructed by using the combination of Gaussian function and adaptive bandwidth.Not all the factors can pass the non-stationary test of time and space.According to the regression results of the model,6 influence factors pass the non-stationary test.Including speed,age,town,body defect,intersection and cloudy.(3)The factors of the severity of motorcycle traffic accidents between young drivers and elderly drivers are quite different in spatial scale and temporal and spatial instability.Among the factors of the severity of motorcycle traffic accidents among young drivers,the spatial scale of the six influence factors including body defects,distracted driving,gender,intersection,speed and left turn is small,and the town have passed the non-stationary test.Among the factors of the severity of motorcycle traffic accidents among elderly drivers,the spatial scale of the four influence factors including drug driving,town,intersection and speed is small,and the intersection,slippery road surface,speed,speed limit,other helmets and no light at night passed the non-stationary test.The research results are helpful to understand the spatial and temporal heterogeneity of the influence factors on the severity of motorcycle traffic accidents.At the same time,it is helpful for the traffic management department to put forward local or global motorcycle traffic safety countermeasures.