Flight Operation Risk Assessment Method Based on BP Neural Network Algorithm

For improving the level of aviation safety management of China, based on BP neural network algorithm, a method for flight operation risk assessment was explored. In this method, B737 and A320 aircrafts were selected as samples, flight scenario during approach and landing phase was constructed by using FDR/QAR data, a risk index system was studied to quantify the risk of specific flight scenario, a set of pilot questionnaire was designed, and a risk model was trained by pilot experiences. The accuracy and reasonability of the estimated risk value were verified by FOQA results of real flights. This method provides a scientific tool for airlines to monitor flight operation risk, and it is helpful for authorities to enhance industrial safety management.