Aircraft Engine Performance Monitoring and Anomaly Detection Using N1 and N1 Parameters

Abstract

Aircraft engine performance monitoring is crucial to ensure flight safety, operational efficiency, and preventive maintenance. This paper presents a comprehensive review and discussion on the application of machine learning techniques to detect anomalies using N1 (fan speed) and N2 (core speed) parameters. These indicators are key for identifying irregularities in engine behavior during different flight conditions. Through a review of recent studies and implementation strategies, this paper demonstrates how data-driven models can improve anomaly detection and support predictive maintenance in modern aviation systems.

Introduction

Modern aircraft engines operate under extreme conditions, requiring continuous monitoring to ensure safety and efficiency. Key parameters for engine health monitoring are N1 and N2, representing the rotational speeds of the low- and high-pressure spools. These metrics reflect engine performance and help detect faults such as compressor stalls or bearing wear.

Various advanced data-driven techniques analyze N1 and N2 time-series data to identify anomalies and predict engine health. Methods include:

• Autoencoders for unsupervised anomaly detection by learning normal behavior patterns.

• LSTM networks for capturing temporal sequences and classifying faults in real-time.

• Hybrid physics-based and neural network models for interpretable and reliable diagnostics.

• Federated learning to train models across multiple aircraft without sharing raw data, ensuring privacy.

• Statistical Process Control (SPC) charts for simple trend monitoring and early fault detection.

• Dimensionality reduction techniques (PCA, ICA) to enhance signal clarity and improve classifier accuracy.

• Digital twins simulating real-time engine behavior for predictive maintenance.

• Ensemble methods (Random Forests, Gradient Boosting) for robust fault classification.

• Kalman filters to smooth sensor data and reduce false alarms.

This paper reviews these methods, highlighting their strengths, challenges, and suitability for integration into real-time aircraft health monitoring systems, aiming to improve predictive maintenance and operational reliability.

Conclusion

Monitoring aircraft engines using N1 and N2 parameters enhances safety and operational reliability. Machine learning models outperform traditional methods in detecting subtle anomalies. Techniques like LSTM, autoencoders, and hybrid models offer real-time insights. These systems reduce unscheduled maintenance and improve decision-making. The approach supports the shift toward intelligent, data-driven aircraft operations. Future models can integrate additional features like EGT, vibration, and oil pressure. Edge computing and federated learning enable onboard, privacy-preserving diagnostics. Cross-platform adaptability is needed for different engines and flight conditions. Explainable AI will improve trust and usability for aviation personnel. Collaboration is key for regulatory approval and industry-wide implementation.

References

[1] Kim, J., et al. “Autoencoder-Based Anomaly Detection for Turbofan Engines.” Journal of Aerospace AI, 2020. [2] Zhao, Y., et al. “LSTM Networks for Aircraft Engine Fault Classification.” Aerospace Systems, 2019. [3] Goebel, K., et al. “Prognostics Framework for Aircraft Engines.” IEEE Aerospace Conference, 2017. [4] Chen, X., and Lee, H. “Hybrid Machine Learning Models for Turbofan Monitoring.” Applied AI Review, 2021. [5] Gao, M., et al. “Federated Learning for Aviation Applications.” Aerospace Machine Intelligence, 2022. [6] Yin, R., et al. “N2-Based Fault Detection Using SPC Charts.” Mechanical Systems and Signal Processing, 2018. [7] Liu, J., et al. “PCA and ICA for Noise Reduction in Engine Diagnostics.” Journal of Data Analytics in Aerospace, 2021. [8] Wang, T., et al. “Digital Twin Models for Predictive Diagnostics in Aviation.” Aerospace Technology Journal, 2019 [9] Zhang, Q., et al. “Ensemble Learning for Engine Fault Detection.” Engineering Applications of Artificial Intelligence, 2020. [10] Tang, S., et al. “Kalman Filtering for N1/N2 Signal Enhancement.” Sensors and Systems Journal, 2020. [11] Saxena, A., and Goebel, K. “The C-MAPSS Dataset for Engine Fault Prognostics.” NASA Technical Report, 2008. [12] Roychoudhury, I., et al. “Survey of Anomaly Detection Techniques for Aircraft Engines.” Journal of Aviation Safety, 2016. [13] Bishop, C. M. “Pattern Recognition and Machine Learning.” Springer, 2006. [14] Lee, D., et al. “Real-Time Decision Support Systems for Engine Monitoring.” Journal of Intelligent Aviation, 2017. [15] [Cheng, Z., et al. “Modeling N2 Decay During Fault Events.” Mechanical Engineering Review, 2014. [16] Liu, F., and Wu, Y. “AI Techniques in Condition-Based Maintenance.” Review of Engineering AI, 2018. [17] ISO 13374. “Condition Monitoring and Diagnostics of Machines.” International Organization for Standardization, 2014. [18]. [18] FAA AC 33.28. “Advisory Circular on Engine Condition Monitoring.” Federal Aviation Administration, 2020. [19] Xie, W., et al. “Clustering of Engine States Using Unsupervised Learning.” Journal of Intelligent Systems, 2015. [20] Pankaj, R., et al. “Jet Engine Predictive Maintenance: A Case Study.” Aerospace Maintenance Review, 2023

Copyright

Copyright © 2025 Sudeep S Shetty, Trupti Shripad Tagare. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.