Multi-Source Data Fusion for Aircraft Structural Health Monitoring: A Review

Multi-source data fusion stands as a pivotal technology for enhancing aircraft structural health monitoring (SHM) system performance and ensuring flight safety. To systematically outline the research landscape and development trajectories in this domain, this paper first elaborates on the theoretical framework of data fusion, encompassing three fusion modalities - competitive, complementary, and collaborative - alongside a three-tiered fusion architecture that spans data-level, feature-level, and decision-level integration. Subsequently, from an application perspective, research advances in data fusion techniques for critical aircraft structures were reviewed. In metallic structures, fusion methodologies have significantly enhanced the detection accuracy of classical damage modes, such as fatigue cracks. For composite structures, multi-modal fusion - particularly when integrated with deep learning - provides powerful tools for addressing complex failure mechanisms, including delamination and impact damage. At full-aircraft level, digital twin-centered fusion platforms are advancing SHM toward systemic predictive health management paradigms. Finally, this paper synthesizes core challenges hindering operational deployment: data heterogeneity, scarcity of damage samples, and poor model interpretability. Future breakthrough directions, represented by physics-informed intelligent fusion, digital twin-driven holographic perception, and Explainable Artificial Intelligence (XAI), were further projected. Collectively, this work delivers a comprehensive knowledge map and forward-looking perspectives for data fusion research and implementation in aircraft SHM systems.