Akademik Veri Yönetim Sistemi
Journal of Thermal Analysis and Calorimetry, cilt.150, sa.18, ss.14169-14181, 2025 (SCI-Expanded, Scopus)
Engine health monitoring, prediction of performance, and diagnostics are key factors in the assessment of gas turbine engines equipped with digital devices. The development of new and more advanced artificial intelligence techniques has significant benefits for airlines in terms of fleet management, cost savings, and ensuring safety obligations. The aim of this study is to examine the performance of PW-4056-3 engine parameters obtained from the digital flight data recorder (DFDR) of the Boeing 747-400F aircraft. The study uses 1D convolutional neural network techniques, training, and testing to predict low- and high-pressure compressors (N1 and N2), exhaust gas temperature (EGT), fuel flow (FF), and vibration (VIB) for selected engines. To prove the study approach is accurate and efficient, different metrics are used, such as mean squared error (MSE), R-squared (R2), and mean absolute error (MAE). The study's results show that, among the selected parameters, the values of N2 remain stable, displaying an average of 94.93%. The EGT temperature maintains around 540 °C with different flight altitudes, which is in accordance with standard values. The FF rate during the same flight phase is in compliance with standard engine parameters, stabilizing the consumption rate after the assigned transition altitude to an average of 6800 kg per hour, demonstrating their highly predictable performance. Regarding the significant variation of the low-pressure compressor's initial anomalies within 81 s after takeoff, the N1 fluctuated between 97 and 97.84%. This fluctuation also significantly influenced the vibration values, revealing abnormal trends with chaotic frequency changes up until the transition level at FL180. The results reveal that the implementation of current methodology has high assumptions about discovering the engine anomalies independent of each other of two parameters, however closely interrelated in the context of their health monitoring, as between N1 and VIB.