深度学习驱动的角联锁机织复合材料低速冲击裂纹网络与损伤演化研究OA

Microcracks in three-dimensional angle interlock woven composites are key factors affect-ing their mechanical properties and structural integrity,yet the systematic analysis of such micro-scopic damages in existing studies is still limited.Automatic extraction and quantitative analysis of microcracks under different low-velocity impact energy conditions were realized based on the U-Net deep learning model.The spatial distribution characteristics of microcracks under different ener-gies were further revealed by introducing a skeleton algorithm to calculate the crack length and width.It is found that the U-Net model has excellent performance in the microcrack recognition task,and the segmentation accuracy reaches 99％.With the elevation of impact energy,the number of mi-crocracks increases significantly.The distribution range expands,and gradually runs through the entire material cross-section in the thickness direction.At the same time,the connectivity between microcracks is enhanced under the high-energy condition,indicating that the material damage evolves toward a more complex networked form.In terms of axial distribution,the microcracks were mainly concentrated in the lower part of the specimen under low-energy impact,while they tended to be uni-formly distributed under high-energy impact.In the weft direction,the microcracks showed stronger concentration and continuity.The above patterns reflect the damage evolution trend of microc-racks from local isolation to overall synergistic development,which provides an important support for the fine characterization and understanding of the impact damage of composites.