Based on the characteristics of QAR as big data, this paper tries to use wavelet transform and autoencoder to pre-process the QAR data, extract the effective features of the data, and further clustering algorithm to identify abnormal QAR data. More than 1,000 sets of QAR data of an airline's Boeing737-800 model were collected, and the feasibility of the method was verified by using real flight data. The results show that this method can detect anomalies in flight data during the landing phase, and the recognition effect is better than directly inputting QAR data for anomaly detection, and can be used as a supplement to traditional anomaly detection methods.
Flight anomaly detection model based on QAR data autoencoder and DBscan algorithm
2021-10-20
1970832 byte
Conference paper
Electronic Resource
English