Aircraft composite structures are widely used due to their advantages such as lightweight and high strength. However, they are prone to hidden damage in complex service environments, making accurate damage localization crucial for flight safety. Traditional damage localization methods rely on single-sensor data, resulting in low localization accuracy and weak anti-interference capabilities, making it difficult to meet the detection requirements of complex structures. This paper first constructs a multi-source sensor data acquisition system, integrating fiber optic gratings, piezoelectric sensors, and infrared thermal imaging data. Second, it proposes a multi-scale feature fusion algorithm based on an attention mechanism to achieve effective fusion of heterogeneous data and extraction of key features. Finally, it builds a damage localization model and introduces an improved particle swarm optimization algorithm to optimize model parameters. The experiment used a carbon fiber composite airfoil simulation as the research object, setting up damage samples of different types, locations, and degrees. The results show that the average training time of the proposed method is 45.2 min, while the traditional multi-source data + D-S evidence theory method is 68.5 min. The improvement is mainly due to the efficient optimization of hyperparameters by the IPSO algorithm, avoiding the blindness of traditional grid search. The time is slightly longer than that of single PZT data + SVM model (32.8 min), because multi-source data processing and deep network models increase computational load, but it is still within an acceptable range, providing an efficient and reliable technical solution for damage localization of aircraft composite structures.
Yang et al. (Thu,) studied this question.