Improved Clustering And Deduction Of Typical Dangerous Scenarios For Passenger Vehicles And Two-wheelers

Two-wheeler traffic accident is one of the main types of urban traffic accidents.The twowheeled cyclists who lack safety protection belong to the weak participants in the road traffic.Once an accident occurs,the two-wheeled cyclists in the collision accident have a higher probability of injury and death.Therefore,this paper excavates and analyzes the accident scenario data of passenger car and two-wheelers,in order to provide an effective construction scheme of typical dangerous scenarios of two-wheeler for intelligent vehicle development and testing.First of all,this paper describes and summarizes the methods of domestic and foreign experts and scholars mining typical dangerous scenarios,and then leads to cluster analysis as the research method of this paper.Then,the National Automobile Accident In-depth Investigation System database(NAIS)is described,including accident case collection standard,accident information collection content,accident content analysis and report,etc.And further sort out some of the collected data to understand the overall data overview,from which 400 cases of passenger car and two-wheelers accidents with complete information collected were selected as the research basis.Secondly,several commonly used cluster analysis methods are briefly introduced,and their characteristics are compared and analyzed.The concrete steps of hierarchical clustering and fuzzy comprehensive evaluation method are introduced in detail.This paper introduces the selection process of static variables in the typical dangerous scenarios of passenger car and twowheelers,as well as the process of supplement and conversion of field content.Statistical analysis of the contents of various fields to understand the distribution of the contents of each field.The basic clustering method is improved by fuzzy comprehensive evaluation method,the concept of "weighted absolute value distance" is introduced,and the mathematical model is established.According to the selected fields,this paper makes an improved cluster analysis of the 400 accident cases,extracts the typical features of each kind of scenario with common characteristics,and finally obtains the typical dangerous scenarios of 7 kinds of passenger cars and two-wheelers.Finally,the distribution of accident cases related to the scenario at the crossroads is statistically analyzed.Taking the second category of the scenario as an example,the dynamic parameters in the scenario,that is,the speed relationship between the two vehicles,are analyzed and discussed,the danger levels are divided,and the hazards of each level are discussed.Thus,the kinematic models of seven kinds of scenarios and the deduction scenarios with the combination of static and dynamic parameters are studied respectively,and finally seven kinds of dangerous scenarios of passenger cars and two-wheelers in accordance with the road traffic conditions in China are obtained.Research conclusion: a new method for constructing typical dangerous scenarios of passenger cars and two-wheelers is proposed in this paper.According to this method,the collected data are processed to get seven kinds of typical dangerous scenarios which accord with the traffic situation of our country,and the dangerous scenario sets are deduced.It can provide theoretical basis for the expansion of road traffic hazard scenario database in the field of active safety,provide two-wheeler test scenario for the early test and development of intelligent driving technology,and improve the test quality and efficiency of intelligent driving vehicles.Enhance the risk response ability of intelligent vehicles,prevent and reduce the occurrence of accidents.