Research on the Flight Anomaly Detection During Take-off Phase Based on FOQA Data
Yunpeng Jiang, Ningning Le, Yufeng Zhang, Yinger Zheng, Yang Jiao · 2019 CAA Symposium on Fault Detection, Supervision and Safety for Technical Processes (SAFEPROCESS) · 2019
Aiming at the high risk of fatal accidents during the flight's take-off phase, a data-driven method for flight anomaly detection is proposed; firstly, the key performance parameters of the take-off phase are chosen; secondly, Airbus A320 aircraft is taken as the research object, and the corresponding QAR data in FOQA station are divided into training set and testing set; thirdly, the training set is clustered by one-class SVM method to get the anomaly detection model, then the model is used to detect the outlier of the testing set; finally, the flight anomaly detection examples of two-dimensional, three-dimensional and multi-dimensional parameters are given respectively, and the outlier flights deviating from the group characteristics can be accurately marked, which provides important decision support information for the flight safety management.