Feature Extraction And Identification Of Aggressive Driving Behaviors Considering Driver’s Propensity

In recent years,with the economic and social development,China’s car ownership has continued to increase.With life’s pace accelerating,people’s pursuit of their own driving interests(including time and space)is getting higher and higher,as the tolerance toward other traffic participants is lowering,and a series of illegal traffic behaviors(such as close-up overtaking,frequent lane changing and not illegal giving way)occur frequently,which will lead to serious traffic accidents and pose a serious threat to people’s life and property.In this study,aggressive driving behaviors in dangerous driving behaviors focuses on,in order to intelligently identify the occurrence of such dangerous driving behaviors,minimize the losses caused by them,and provide theoretical and technical support for the development of automotive active safety early warning systems.The main research work as follows:Firstly,the research status on aggressive behaviors at home and abroad was reviewed,and the background significance of this study was clarified,in addition,the mechanism and influencing factors were systematically learned.On this basis,the current research deficiencies and the research content and technical route of this paper are clarified.Secondly,based on the causes of aggressive driving behaviors,the extension of driver’s propensity and the characteristics of aggressive behaviors,using behavior investigation,field investigation,reliability and validity analysis methods,the preparation and judgment of driver’s propensity questionnaire were completed.Thirdly,the experiment of aggressive driving behaviors were carried out,and the multi-sensor data of the physiology-psychological,vehicle operation and environmental information of the driver during the driving process was collected.Using the data mining technology and intelligent algorithm,the multi-dimensional sensor anomaly detection model was established and completed the data preprocessing work.Forth,according to the data characteristics of this paper,in the data dimension and computational complexity,and improve the subsequent identification efficiency,the comparison verification and simulation results analysis of different feature extraction models were carried out.Finally,identification of aggressive behaviors have been completed and the preliminary ideas for design of next early warning system have been provided.And then,the deep neural network identification models of different aggressive driving behaviors were built by Matlab toolbox and Python platform based on the basic models of different neural networks,and the invasive driving behaviors recognition and analysis of drivers with different driver’s propensity types were completed,and a preliminary idea for precise aggressive driving behaviors early warning system is put forward.