Study On GA-BP Prediction Method For The Severity Of Traffic Accidents On Expressway

In recent years,with the growth of vehicle index in China,the contradiction between supply and demand between traffic resources and vehicles is becoming increasingly serious,and traffic accidents occur frequently.People from all walks of life pay close attention to this problem and are eager to solve it.With the rapid development of artificial intelligence and the guidance of national policies,it is of great significance to study and construct the intelligent prediction method of traffic accidents for the development of intelligent monitoring and forecasting,early warning and processing of traffic accidents.In this paper,based on the current situation of highway traffic safety,firstly,the existing research results are analyzed from the analysis of accident influencing factors and the intelligent prediction method of traffic accidents,and BP neural network and genetic algorithm are used to intelligently predict the highway traffic accident severity.Secondly,the definition of traffic accidents and the basis for dividing the traffic accident severity in China are determined.Based on the interaction mechanism of the four factors of people,vehicle,road and environment in the traffic system and the historical data of traffic accidents in Yunnan Province K highway,the influence factors of traffic accident severity in highway are analyzed comprehensively and systematically,and the preliminary indexes are quantified by using extreme value statistics,and the characteristic indexes are simplified and selected by using principal component analysis.Then,based on the characteristics and shortcomings of neural network and genetic algorithm,the genetic algorithm are put forward to optimize the weight and threshold of BP neural network structure to model,the parameter design and modeling process are gave in detail,and the GA-BP neural network prediction model of highway traffic accidents are realized in MATLAB.Finally,the intelligent prediction method of traffic accidents is verified by analyzing data of K highway in Yunnan Province.After processing and classifying the accident data,312 accident data are used to design and learn the basic structure of the model,and 100 accident data are used to verify the model.The results show that the predicted value of the model is basically consistent with the actual accident severity,and the prediction accuracy is as high as 95%.The mean square error of BP neural network optimized by genetic algorithm is reduced from 1.89×10-4 to 5.37×10-5.The research results in this paper can be used for the real-time monitoring and prediction of traffic accidents in the later period,which is convenient for the early warning,timely intervention,optimization and treatment of traffic accidents.At the same time,it lays the foundation for the realization of the intelligent systems to intelligently monitor,predict,early warn and treat the traffic accidents.