Identification And Hazard Degree Evaluation Of Bus Driving Behavior Based On Multi-source Data Fusion

Bus drivers are the dual factors that affect the quality of public transport service and service satisfaction.Their personal behavior reflects the overall image of the city and urban tolerance.However,in the development of modern public transport,some bad driving behaviors have become the biggest constraints that affect the safety of public transport operations and improve the quality of public transport services,and have also become a hot issue of concern to the public service industry around the world.From the perspective of traffic behavior and traffic safety,this paper discusses the identification of bus driving behavior and the evaluation of hazard degree.This paper mainly focuses on BP Ada Boost model,D-S evidence theory,fuzzy evaluation and other methods to build the evaluation system of driving behavior recognition.The main work fully applied in this paper includes:(1)A bus driving behavior recognition model based on BP Ada Boost and GPS data feature analysis is proposed;Identify and analyze the characteristics of GPS driving behavior.Firstly,the GPS driving behavior data of XX city is deeply analyzed and preprocessed;Then extract driving behavior characteristics;Finally,the data characteristics obtained from the analysis are used to identify and verify the model,and the recognition rate of driving behavior is 93.67%.(2)A bus driving behavior recognition method based on vehicle video data fusion is proposed.By analyzing the characteristics of on-board driving video data,BP Ada Boost model is also established to identify violations,and the recognition rate of driving behavior is93.40%.(3)Based on fuzzy evaluation and D-S evidence theory,a method to evaluate the harm degree of driving behavior is proposed.This paper further discusses the principles of fuzzy evaluation and D-S evidence theory methods,and deeply applies these methods.On this basis,combined with the advantages of the two methods,we can further analyze the driving behavior data,which can accurately evaluate the harm degree of abnormal driver behavior,so that the examples can be effectively verified.