Research On The Relationship Between Driver 's Risk Perception And Driving Behavior

According to statistics of traffic safety accidents from domestic and foreign, human factors are the main factors of traffic accidents. The driver’s ability of risk perception has a direct impact on the safety of driving, therefore, research on risk perception characteristics of the driver has important theoretical and practical significance.Take the effectiveness of driver risk perception as the breakthrough point, this paper applies KMRTDS driving simulation platform to study the relationship between risk perception characteristics of the driver and driving behavior. Based on the experimental data of risk perception characteristics, clustering analysis can make the results of the driver’s types, analysis the effectiveness between risk perception and driving behavior parameters, build a forecasting model between factors of driving characteristics and type of drivers Details are as follows:Recruitment of 61 drivers as subjects participated in risk perception characteristics testing laboratories, driving simulation system, real-time recording the subjects of driving behavior data and vehicle operating state data; perceived effectiveness based on traffic conflict analysis theory and risk, calculate risk subjects perceived utility value, using cluster analysis method will be divided into four kinds of test driver perceived risk types; extraction 7 indicators in driving behavior and 28 parameters from the vehicle running state data, qualitative analysis of the effectiveness of risk perception and each parameter trends; through correlation analysis, the mean longitudinal speed, longitudinal velocity maximum 12 parameters and risk perception utility has significant correlation; at the same time, for the information overlap parameters collinearity caused by principal component analysis excluding joint degrees lower parameters, finally get five effective characterization of driving behavior characteristic independent factor.BP artificial neural network-based approach to construct five driving behavior characteristic risk factor as the driver input sensing type prediction models. Longitudinal model shows that the driver’s vehicle speed is an important indicator to characterize the type of driver; lateral movement of the vehicle longitudinal comparing features, the better characterization of the driver’s risk perception characteristics. Model consistent with the actual conclusion, the research results of this project for the driver’s risk perception research features provide a theoretical basis and methods for reference.