Abstract: Aiming at the challenge of monitoring fatigue cracks in spacecraft structures during service, an intelligent identification method for fatigue cracks based on fiber Bragg gratings and ultrasonic guided waves was proposed in this paper. Firstly, a hybrid approach combining passive monitoring using fiber sensing with active detection via ultrasonic guided waves was employed to comprehensively acquire the state data of spacecraft structures. Then, a deep learning network of stacked denoising autoencoders was adopted to construct a fatigue crack identification model for structures, which could adaptively extract crack damage features directly from the structural state data and represent the complex mapping relationship between the structural states and fatigue cracks with a deep model, achieving precise identification of crack locations and lengths. Experimental results show that the identification accuracy of fatigue cracks in the proposed method is more than 90%, which can meet the application requirement of autonomous identification of fatigue cracks in on-orbit spacecraft structures.

Key words: structural health monitoring, fatigue crack, intelligent identification, optical fiber sensing, ultrasonic guided wave