The Feasibility Analysis Of Military Vehicles Accident Prediction Based BP Neural Network

With the development of China's national defense transportation and the growing of a variety of military vehicles, military vehicle accidents have occurred occasionally. Making accurate and effective military vehicle traffic predction plays an important role for consolidation of national defense and improving combat effectiveness.At the present stage, military vehicle accident research methods is limited to summarizing the experience, and lacking the method of quantification.This paper shows that BP neural network can predict the probability of an accident accurately and effectively, and could reduce the loss of property and personnel.Main contentsGenerally overviews the background, the purpose and significance, methods and international status of the thesis, etcExplains the principle of BP neural network to predict the military vehicle accidents. BP neural network is consist of a large number of processing units, has a strong parallel computing,learning,fault tolerance and anti-jamming capability, tames the network and establishes the model by the input vector and the target output. The paper adopts the method of additional momentum to overcome the slow in convergence and falling into local minimum easily.Analyses the causes of accident from seven aspects, summed up the quality of the driver, a safety management system, driving experience, the average daily mileage, education, environment and vehicle condition. Bulids up BP neural network model by way of questionnaires.Prove that the BP neural network is feasible and valuable for the prediction of military vehicle accidents through the establishment of the test model.