Abstract: In view of that using a single algorithm is difficult to greatly improve the accuracy of flight operation risk identification,this article analyzes the flight operation workflow in detail first.Fifteen important evaluation indicators were screened out.A hundred flight cases of X Airlines were selected as data samples.Nine risk core indicators were got by using rough set theory reduction.Then,two kinds of machine learning methods,SVM and neural network,were used to build a risk identification model for calculating the risk levels,and a comparison was made between the levels and the operation result of X-Air control system based on fuzzy algorithm.A multi-algorithm collaboration model was built according to the advantages and disadvantages of each algorithm.Finally,operational data of G-Air and N-Air was used to test the applicability of the model.The results show that the neural network method has the best effect on low-risk discrimination,SVM has the strongest ability to identify medium and high risks,and a correct rate up to 95% can be obtained by using the multi-algorithm collaboration model.

Key words: flight operations, neural network algorithm, support vector machine(SVM), rough sets, multi-algorithm collaboration