The Research Of Road Traffic Accident Predictive Model Based On Genetic BP Neural Networks

The road traffic safety is a big complicated system, which involves the politics, economy, culture, technique and so on. And the road traffic accidents indicate concretely the level of road traffic safety. China is one of the countries that have most road traffic accidents in the world, and the death toll of traffic accidents occupies the first place of the world. So road traffic safety situation is very serious at present.The traffic accidents forecast and prevention is the main means to improve the level of road traffic safety. Contrasted the national road traffic situation with western developed country, this paper analyzed traffic accidents cause analysis theory, studied theory and methods of macroscopic and the microscopic forecast model, built the application of road traffic safety analysis based on genetic BP neural networks. The application can analyze and forecast road traffic accident thoroughly, scientifically and systemically.Firstly, contrasted and analyzed the typical accidents cause theories, this paper studied the cause and effect theory model to analyze road traffic accidents. Furthermore, the theory was used to study the unsafe system factors of traffic accidents occurs on Hu-Hang-Yong and Hang-Jin-Qu freeway in the charge of Hangzhou traffic police's precinct. Then safety measures are brought forward to reduce freeway accidents.Secondly, in the forecast methods aspect, the BP neural networks model based on genetic algorithm was used to forecast macroscopic traffic accidents. It can grasp the safety trend transformation of a country or a region. Then scientific gist can be provided for road safety administrator institution to constitute traffic safety policy. The result indicated that this model can well predict and had good capability to extend network, and the conclusion was satisfying.Finally, in view of the specific microscopic question of road traffic accident prevention, microscopic forecast model base on the genetic BP neural networks was established. This paper gave an example to validate the model, which used the fact road structure and accident data of Hu-Hang-Yong and Hang-Jin-Qu freeway in the charge of Hangzhou traffic police's precinct. The result indicated that this model can well predict road traffic accidents.