AI-driven structural health monitoring for UAVs: a review of damage detection and safety enhancement

This review examines AI-driven structural health monitoring (SHM) for unmanned aerial vehicles (UAVs), focusing on damage detection and safety enhancement. UAVs face structural risks from fatigue, environmental exposure, and propulsion failures that conventional inspection methods like visual checks or ultrasonic testing, often miss early. Recent research introduces AI-based frameworks integrating multisensor data for real-time monitoring and predictive diagnostics. The study explores model-based and data-driven approaches, including convolutional neural networks (CNNs), long short-term memory networks (LSTMs), autoencoders, and physics-informed models. These algorithms identify anomalies in vibration, acoustic, thermal, and current signals, improving fault detection accuracy. Comparative analysis reveals conventional SHM methods provide baseline reliability and regulatory compliance, while AI-driven solutions offer greater adaptability, scalability, and suitability for in-flight assessments capabilities. Together, they form a hybrid inspection ecosystem enhancing operational safety and reducing maintenance demands. Future research should focus on lightweight embedded implementations, explainable AI techniques, and UAV-specific failure datasets. Advances in sensor fusion, edge computing, and semi-supervised learning are critical to overcoming current barriers, including limited data and computational constraints. By combining conventional inspection methods with intelligent AI architectures, UAV SHM can shift from post-flight detection to proactive fault prevention, thereby improving mission safety and reliability.