The Research Of The Aviation Operation Risk Pre-Warning Model Based On Machine Learning

With the rapid development of the civil aviation industry,the surge in air traffic volume and the enhancement in aircraft design requirements have led to an exponential growth in the amount and the complexity of the aviation operational risk.To a certain extent,the experience-based qualitative risk analysis method can effectively guarantee the aviation operational safety,but its own limitations restrict the further improvement of the operational safety level.How to use data technology to establish a more forward-looking and more accurate aviation operational risk management and pre-warning system is an urgent problem to be solved during the current civil aviation development.This paper proposes an aviation operational risk pre-warning model based on machine learning theories.An aviation operational risk pre-warning parameter identification method is established.Through the mechanism analysis,the event correlation elements are extracted,the pre-warning logic is established,and combined with the QAR parameter set,the risk pre-warning parameters of the certain aviation operational event are identified.The data processing methods of aviation operation sample are studied.Based on the analysis on aviation operational event determination and risk interval division,a method for determining the aviation operational sample risk category of is proposed.According to the parameter data characteristics,the vacancy value filling methods for the risk pre-warning parameter are given,such as the polynomial curve fitting method and the genetic optimization BP neural network method.Based on the support vector machine classification theory,aviation operation risk classifiers based on genetic optimization and adaptive mutation particle swarm optimization are established,and the aviation operation risk pre-warning model is built,the risk pre-warning function of which can be realized by monitoring and analyzing the real-time data of risk warning parameters.Two examples,heavy landing and APU-IGV fracture,are given to verify the effectiveness of the risk warning model.