Intelligent composite structures with self-aware functions are preferable for future aircrafts. The real-time location of damaged areas of composites is a key step. In this study, deep transfer learning was used to achieve the real-time location of damaged areas. The sensor network obtained acoustic emission signals from different damaged areas of the aluminum alloy plate. The acoustic emission time-domain signal is transformed into the input image by continuous wavelet transform. The convolutional neural network-based model automatically localized the damaged area by extracting features from the input image. A small amount of composite acoustic emission data was used to fine-tune some network parameters of the basic model through transfer learning. This enabled the model to classify the damaged area of composites. The accuracy of the transfer learning model trained with 900 samples is 96.38%, which is comparable to the accuracy of the model trained directly with 1800 samples; the training time of the former is only 17.68% of that of the latter. The proposed method can be easily adapted to new composite structures using transfer learning and a small dataset, providing a new idea for structural health monitoring.
This study proposes a prediction method for residual compressive strength after impact based on the extreme gradient boosting model, focusing on composite laminates as the studied material system. Acoustic emission tests were conducted under controlled temperature and humidity conditions to collect characteristic parameters, establishing a mapping relationship between these parameters and residual compressive strength under small sample conditions. The model accurately predicted the residual compressive strength of the laminates after impact, with the coefficient of determination and root mean square error for the test set being 0.9910 and 2.9174, respectively. A comparison of the performance of the artificial neural network model and the extreme gradient boosting model shows that, in the case of small data volumes, the extreme gradient boosting model exhibits superior accuracy and robustness compared to the artificial neural network. Furthermore, the sensitivity of acoustic emission characteristic parameters is analyzed using the SHAP method, revealing that indicators such as peak amplitude, ring count, energy, and peak frequency significantly impact the prediction results of residual compressive strength. The machine-learning-based method for assessing the damage tolerance of composite laminates proposed in this paper utilizes the global monitoring advantages of acoustic emission technology to rapidly predict the residual compressive strength after the impact of composite laminates, providing a theoretical approach for online structural health monitoring of composite laminates. This method is applicable to various composite laminate structures under different impact conditions, demonstrating its broad applicability and reliability.
This study proposes a bionic helical configuration design concept, focusing on glass-fiber-reinforced polymer matrix composites. Through a combination of experimental and numerical simulation methods, it systematically investigates the resistance to multiple impacts and damage tolerance. The research designs and fabricates two types of bionic laminates: a cross-helical and a symmetric-helical structures. By conducting repeated impact experiments at 5 J of energy for 1, 5, 10, and 15 impact times and employing advanced characterization techniques, such as ultrasonic C-scan and X-ray CT, the study reveals the mechanisms of interlaminar damage propagation and failure characteristics. Based on experimental findings, a finite element model encompassing the entire impact process and post-impact compression behavior is established. Utilizing this model, three optimized novel bionic configurations are further developed, providing new insights and theoretical support for the structural design of high-performance impact-resistant polymer matrix composites.