Advanced AI-Driven Architecture for Real-Time Monitoring and Intelligent Fault Detection of Aircraft Engine Compressor and Fuel Systems under Emergency Operating Conditions

Abstract

The safety and operational efficiency of aircraft engines depend heavily on the ability to continuously monitor critical components, detect incipient faults, and respond rapidly to emergency conditions. Among the most safety-critical subsystems, the compressor and fuel system play a pivotal role in maintaining optimal thrust and engine performance; their failure during flight can lead to catastrophic consequences. This study proposes an advanced AI-driven architecture specifically designed for real-time monitoring and intelligent fault detection of these systems under emergency operating conditions. The framework integrates high-resolution sensor networks with advanced signal processing techniques to capture and preprocess multidimensional operational data, including pressure, temperature, vibration, and fuel flow parameters. A hybrid machine learning approach combining feature engineering, deep neural networks, and ensemble classifiers is employed to identify anomalies, diagnose fault types, and predict potential failure progression. To enhance accuracy and adaptability, the architecture incorporates a digital twin of the compressor and fuel systems, enabling the simulation of diverse fault scenarios, optimization of detection thresholds, and continuous model learning. The system’s layered design consists of a data acquisition layer for high-frequency signal collection, a diagnostic layer for intelligent decision-making, and an emergency response layer that supports flight control systems in initiating corrective measures, including thrust modulation and safe landing protocols. Performance evaluation is conducted using both simulated datasets from the digital twin model and real-world engine operational data obtained from test benches and flight records. Results demonstrate that the proposed system significantly outperforms conventional fault detection and diagnostic methods, achieving detection accuracies above 98%, reducing false alarm rates by over 40%, and enabling fault localization within milliseconds of anomaly onset. Furthermore, the architecture’s predictive capabilities facilitate proactive emergency management, minimizing the risk of in-flight engine failure and providing pilots with actionable insights to ensure safe landing procedures. This research establishes a robust, scalable, and adaptive framework that advances the state of the art in aircraft engine health monitoring. The proposed methodology offers significant implications for predictive maintenance, operational reliability, and the design of next-generation mechanical-aerospace safety systems, aligning with the evolving needs of Industry 4.0 and AI-enabled aviation technologies.