In modern years, the aviation industry has increasingly depended on Automatic Dependent SurveillanceBroadcast (ADS-B) [3] [4] dataset for observing and maintaining aircraft operations. This paper proposes a hybrid approach that associated Fuzzy Logic and Long Short-Term Memory (LSTM) networks to improved flight extraction and phase identification from large ADS-B datasets. The proposed methodology shows the strengths of Fuzzy Logic in handling uncertainty and opacity in flight dynamics, with the capabilities of LSTM networks for sequential data analysis. This envision some advanced processing large-scale ADS-B datasets, implementing C-Means clustering for initial flight path segmentation, and integrating fuzzy logic with LSTM predictions to elaborate phase classification. This hybrid approach is validated with 92 % accuracy, by using real flight data, showing its usefulness in distinguishing flight phases as ground, climb, descent, cruise, and level. This method shows the importance of aviation data analysis which can be greatly use for flight phase classification, leads to safety and operational decision making of flights.