A Deep Compression and Feature Extraction Method for Flight Performance Analysis of Time Series Aeronautical Data

In order to improve the efficiency and accuracy of flight performance analysis, this paper studies the QAR data of Cessna 172R aircraft and proposes a deep compression algorithm based on adaptive segmented prediction coding and residual quantization, integrated with a multi-dimensional feature extraction framework. Compared to traditional methods, the proposed approach achieves an 8.7:1 compression ratio, while maintaining over 99% extraction accuracy for key flight and engine parameters and delivering stable comprehensive performance scores. Experimental results demonstrate that this method significantly enhances storage and analysis efficiency, ensures data integrity, and provides reliable support for flight quality evaluation and engine condition monitoring.