Anomaly detection of flight states in commercial jet aircraft based on parameter density clustering

Flight safety has always been a critical issue in commercial aviation. In response to the massive, unlabeled parameter data generated by commercial jet aircraft, this paper proposes a flight state anomaly detection method based on parameter density clustering, aimed at improving the efficiency and accuracy of flight safety monitoring for commercial jets, and enhancing the reliability of the aviation industry. This method applies unsupervised analysis of flight parameter data through parameter density clustering to identify and prevent potential anomalous behaviors. First, the data is preprocessed using a sliding window consistency method to align landing time points. Then, Pearson correlation analysis is used for dimensionality reduction to reduce parameter coupling. Subsequently, the parameter density clustering algorithm is applied to the filtered parameter groups to perform density-based clustering and identify anomalous time points. Finally, statistical analysis is conducted to compare the anomalous and normal flight sequences, thereby validating the effectiveness of the method. Experimental results show that this approach can effectively detect anomalies in flight data, providing strong support for flight safety.