Segments Based On Bp Neural Network Security Assessment Study

Civil aviation is a high-risk industry, flight safety is highly related to the safety of passengers and security of property, and even it is a hot topic that people often discuss and study. How to reduce accidents becomes the major task of the civil aviation in China and even around the world. Aviation accidents have become basic elements in restriction and influence the development of the civil aviation, a effective way to deal with it is to prevent in advance. Therefore, establishing a comprehensive and systematic flight-phase safety system has become the most pressing things to enhance civil aviation safety management.Several flight safety theoretical models such as Murphy's theorem and Dominoes theory, which are currently mainly used, make the boundary between each error in causing the accidents very clear and it also can analyze each error specifically. However, they can not lead to a satisfied result in evaluation the safety risk for the whole flight; flight risk can not be quantified. This is the problem that this research is going to solve.This study is going to establish a flight-phase safety risk evaluation framework which is based on the BP Neural network framework, because the neural network technology has better solutions for modeling issues in complex system in terms of uncertainty, serious nonlinear and time-varying. Model used in this research can get safety margin for certain height in the flight based on the analysis of certain scene parameters and evaluate the risk for the whole flight. Accurate risk evaluation not only can identify flight in high risk but also enhance the sense of the pilot so that proper mitigation measure can be made. Moreover, as a effective tool in analyzing risk for airline companies and security department in government, it can be used to enhance the level of civil aviation safety management.Finally, some problems will be given out for further study.