Exhaust Gas Temperature (EGT) serves as a vital indicator in jet engine health monitoring, particularly during high-power phases such as takeoff. Accurate prediction and monitoring of EGT can play a crucial role in ensuring engine efficiency, preventing overheating, and supporting predictive maintenance strategies. This review explores various data-driven approaches for EGT prediction, emphasizing the application of machine learning techniques such as linear regression. The integration of flight parameters like thrust and airspeed enables more accurate estimations of EGT behavior under dynamic conditions. In addition, visualization tools such as Google Charts and Python-based data analysis platforms (e.g., pandas, scikit-learn, and matplotlib) are discussed for their effectiveness in presenting real-time predictions and alerts. The paper also highlights how web-based simulations and alert systems can improve the interpretability and responsiveness of EGT monitoring frameworks. This review aims to provide a comprehensive understanding of EGT prediction technologies and their implications for enhancing modern aircraft engine diagnostics.
Introduction
Monitoring aircraft engine performance, especially Exhaust Gas Temperature (EGT), is crucial for safety and maintenance, as excessive EGT can cause engine damage or failure. Traditionally, EGT has been monitored via threshold-based systems, but advances in real-time data and predictive analytics—particularly machine learning—enable more accurate forecasting of EGT behavior. This paper reviews current prediction techniques, focusing on multivariate linear regression models that use key flight parameters like thrust and airspeed. It also highlights the use of Python for model building and Google Charts for interactive visualization to improve interpretability and real-time monitoring.
The background explains how jet engines produce thrust and why controlling EGT during high-power phases like takeoff is vital. The paper advocates for data-driven predictive models over static threshold methods to enable proactive maintenance and reduce unscheduled downtime. The methodology involves collecting flight data, selecting relevant features, training regression models with Python libraries, and visualizing actual versus predicted EGT values dynamically with alert systems.
The literature review surveys multiple advanced approaches to EGT prediction, including hybrid physical and data-driven models, deep learning architectures like LSTM networks enhanced with correction methods, symbolic regression for interpretability, and ensemble models like LightGBM tuned via metaheuristics. Other methods discussed include deep convolutional neural networks combined with boosting models, generalized regression neural networks, genetic programming, and fusion models (CNN-LSTM) for capturing complex temporal and spatial dependencies. These studies demonstrate continuous improvements in prediction accuracy, robustness, and real-time applicability, supporting the development of interpretable and efficient EGT monitoring systems.
Conclusion
In this review, we have examined the state-of-the-art methodologies for predicting Exhaust Gas Temperature (EGT) at takeoff, emphasizing the application of machine learning techniques, hybrid models, and data-driven approaches. The combination of real-world flight data with theoretical models has shown significant promise in improving prediction accuracy and realtime application capabilities. Techniques like Long Short-Term Memory (LSTM) networks, Symbolic Regression (SR), and advanced ensemble methods such as LightGBM have demonstrated their effectiveness in addressing the complex, nonlinear nature of EGT dynamics. The findings suggest that while current models have made substantial progress in terms of accuracy and adaptability, there is still room for improvement. For instance, the integration of additional variables, such as environmental factors (e.g., weather conditions and altitude) and more granular sensor data, could further enhance the robustness and reliability of predictions.
References
[1] K. Sharma and P. Banerjee, \"Development of a Machine Learning Model for Predicting Abnormalities of Commercial Airplanes,\" Data Science and Management, vol. 7, pp. 256–265, 2024. [Online]. Available: ScienceDirect.
[2] H. Zhang and W. Li, \"Application of Physical-Structure-Driven Deep Learning and Compensation Methods in Aircraft Engine Health Management,\" Engineering Applications of AI, vol. 57, pp. 10–23, 2024. [Online]. Available: Elsevier.
[3] M. Lee, \"Explainable Artificial Intelligence for Exhaust Gas Temperature of Turbofan Engines,\" Preprint (arXiv), 2022. [Online]. Available: arXiv.
[4] J. Huang and L. Chen, \"Aero-engine EGT Prediction Based on LightGBM Optimized by Improved Bat Algorithm,\" Conference paper, date NA. [Online]. Available: IEEE Xplore.
[5] S. Zhang, \"Aero-engine EGT Prediction Based on DCNN and LightGBM Combination,\" 41st Chinese Control Conference, 2022. [Online]. Available: IEEE Xplore.
[6] X. Li and Y. Wang, \"Data-Driven Exhaust Gas Temperature Baseline Predictions for Aeroengine,\" Aerospace, vol. 118, pp. 54–66, 2023. [Online]. Available: MDPI.
[7] L. Yang and J. Zhao, \"Prediction of Jet Engine Parameters Using Genetic Programming,\" UKSim-AMSS Conference, 2014. [Online]. Available: IEEE Xplore.
[8] R. Kumar and M. Patel, \"Performance Sensing Data Prediction for Aircraft APU Using Optimized ELM,\" Sensors, vol. 19, pp. 1235–1247, 2019. [Online]. Available: MDPI.
[9] X. Zhang and C. Feng, \"Study on Advanced EGT Prediction Model Based on LSTM,\" Chinese Journal Paper, 2023. [Online]. Available: CNKI.
[10] Z. Wang and Y. Luo, \"EGT Prediction Based on CNN-LSTM Fusion,\" Chinese Internal Publication, 2022. [Online]. Available: CNKI.
[11] L. Zhao and H. Zhang, \"Improved Machine Learning Models for Engine Exhaust Temperature Prediction,\" Journal of Aerospace Engineering, vol. 35, pp. 128–140, 2020. [Online]. Available: ScienceDirect.
[12] Y. Wang and J. Liu, \"A Hybrid Deep Learning Approach for EGT Prediction in Aircraft Engines,\" International Journal of Aeronautical and Space Sciences, vol. 22, pp. 45–58, 2021. [Online]. Available: Springer.
[13] Q. Chen and X. Li, \"Data-Driven Approach to Predicting Exhaust Gas Temperature in Turbofan Engines,\" Aerospace Science and Technology, vol. 116, p. 105987, 2022. [Online]. Available: Elsevier.
[14] F. Deng and R. Yang, \"Real-Time Engine Health Monitoring Using Machine Learning,\" AIAA Journal of Aircraft, vol. 56, pp. 2899–2910, 2019. [Online]. Available: AIAA.
[15] R. Singh and M. Patel, \"A Comprehensive Review of Predictive Maintenance Techniques for Aircraft Engines,\" Journal of Mechanical Engineering, vol. 68, pp. 2347–2365, 2020. [Online]. Available: Wiley.
[16] R. Kumar and V. Sharma, \"Predicting Exhaust Gas Temperature in Real-Time with Ensemble Machine Learning Models,\" Journal of Aerospace Systems, vol. 30, pp. 1213–1224, 2021. [Online]. Available: Springer.
[17] X. Bai and Y. Sun, \"Artificial Intelligence for Aircraft Engine Diagnostics and Prognostics,\" Computers in Industry, vol. 133, p. 103560, 2022. [Online]. Available: Elsevier.
[18] Z. Zhang and C. Xu, \"Hybrid Modeling for Engine Performance Prediction: A Case Study on EGT,\" Aerospace Engineering and Technology, vol. 34, pp. 561–575, 2021. [Online]. Available: Elsevier.
[19] X. Feng and D. Wang, \"Long-Term Forecasting of Engine Exhaust Gas Temperature Using Time Series Analysis,\" Journal of Aerospace Engineering, vol. 42, pp. 304–315, 2023. [Online]. Available: Wiley.
[20] T. Li and W. Zhang, \"Deep Learning Applications in Aircraft Engine Performance Prediction,\" Applied Soft Computing, vol. 91, p. 106229, 2020. [Online]. Available: Elsevier.