纤维增强复合材料（Fiber-reinforced composites, FRC）因具有极高的比强度和比模量在近几十年来备受关注，被广泛应用于各个领域。结合其舒适性、可设计的优势，更是在冲击防护领域得到了快速地发展。纤维复合材料防护装备的纤维体积分数一般在80%以上。因此，对纤维材料的深入研究是防护装备设计和制造的基础，具有重要意义。本文主要围绕高性能纤维单丝材料、纤维/树脂界面和FRC在高应变率加载下的力学响应以及多尺度表征进行了相关研究。建立了针对微米直径纤维单丝冲击加载实验技术，为微尺度纤维单丝及界面动态力学性能表征提供了有效的实验手段，系统揭示了弹道侵彻条件下纤维单丝及基体的耗能机制；发展了曲面纤维增强复合材料抗侵彻性能分析方法，获得了防护性能的主控参数及其影响规律，为曲面防弹复合材料设计与制备提供了基本依据；初步建立了基于机器学习的高性能纤维增强复合材料侵彻高效计算模型，为防弹复合材料性能预测和高效设计提供了新的途径。主要研究内容包括：

1.针对纤维单丝的动态拉伸性能，改进了微型霍普金森杆实验装置，使用高灵敏度的压电式力传感器代替透射杆，并在入射杆端部增加吸收长杆，使其可以对微米直径纤维单丝的动态力学性能进行高效、准确地测量，并结合扫描电子显微镜观察到的形貌特征，解释了拉伸强度的应变率敏感性及机制。

2.针对纤维单丝受高速横向冲击载荷下的动力学行为，从纤维单丝吸能的角度出发，对物理过程进行了理论推导，获得了横向冲击吸能与纤维单丝横波波速间的定量关系，提出了纤维材料弹道防护性能的评价指标。在此基础上，改进了激光驱动微颗粒冲击实验装置，建立了激光驱动微锥形变形体对微米直径纤维单丝高速横向冲击实验方法，通过测量图像中锥形变形体速度和横波波速来评估纤维单丝在高速横向冲击下的耗能行为。

3.使用多胺改性的方法对纤维表面进行处理，得到了改性对于纤维表面的元素及化学键的影响规律。进一步使用改进的微型霍普金森拉杆对界面的动态力学性能进行了测试，得到了界面剪切强度和刚度等参数。在此基础上，建立了细观代表性体积单元，研究了改性方法、纤维含量和树脂不均匀性对于宏观单层板性能的影响。

4.使用量纲分析给出了FRC抗侵彻性能的主控因素，并建立了复合材料侵彻宏观计算模型，探究了无量纲曲率半径和无量纲面密度对弹道极限速度的影响规律，给出了经验计算公式。并在此基础上研究了复合材料铺层角度对于抗侵彻性能的影响规律。

5.使用机器学习方法研究了复合材料微结构与复合材料抗侵彻性能之间的关联。通过数值模拟建立了机器学习模型数据库，获得了降阶后的数据库主要特征，分析了不同算法及不同参数对于预测模型的影响，明确了预测纤维增强复合材料防护性能的有效预测模型和参数。

In recent decades, fiber-reinforced composites (FRC) have gained significant attention due to their high specific strength and modulus, making them suitable for a wide range of applications. Especially, they are widely used in impact protective engineering due to their comfort and designability. Protective equipment typically utilizes FRCs with a fiber volume fraction above 80%, making an in-depth study of fiber materials crucial for their design and manufacture. This dissertation focuses on several aspects of FRCs under high strain rate loadings, including the mechanical response of high-performance monofilament, the fiber/resin interface, and the characterization methods used in related research. The study establishes an experimental technique for impact loading of micron-diameter monofilament, providing an effective way to characterize the dynamic mechanical properties of micro and nano-scale monofilaments and interfaces. The energy dissipation mechanism of monofilament and matrix under ballistic penetration conditions is systematically revealed. Additionally, the study develops an analysis method for the penetration resistance of curved FRCs, obtaining the vital parameters of protection performance and their scaling laws, providing a basis for the design and preparation of curved ballistic composites. Finally, a preliminary machine learning-based high-performance fiber reinforced composites penetration efficient calculation model is established, providing a new way for prediction and efficient design of bulletproof composites performance. Overall, the main research contents of this dissertation include:

1. To investigate the dynamic tensile properties of monofilaments, we improved a miniature Hopkinson bar device by using a highly sensitive piezoelectric force sensor instead of a transmitted bar, and by adding an absorbing bar at the end of the incident bar. This allowed for efficient and accurate measurement of the dynamic mechanical properties of micron-scale monofilaments. By combining these results with morphological features observed through scanning electron microscopy, we were able to explain the strain-rate sensitivity of their tensile strength.

2. In order to better understand the behavior of monofilaments under high-speed transverse impact loading, we analyzed theoretically the energy absorption behavior of the monofilament and proposed an indicator to estimate the anti-ballistic impact capability. Then improved a laser-driven microparticle impact experimental setup to use a laser-driven slug-like conical deformer, which allowed us to evaluate the energy dissipation behavior of monofilaments under high-speed transverse impact by measuring the conical deformer velocity and transverse wave velocity in images.

3. Using a polyamine modification method, we modified fibers and obtained the effect of the modification on the elemental and chemical bonding patterns of the fiber surface. Then tested the dynamic mechanical properties of the interface using a modified Mini-SHTB, and obtained parameters such as interface shear strength and stiffness. Using this data, we investigated the effects of modification, fiber content, and resin inhomogeneity on the properties of macroscopic monolayers by establishing fine-scale representative volume cells.

4. We discussed the main controlling factors of the ballistic impact resistance of FRC using the dimensional analysis method and established a scaling law. A macroscopic model is also built to investigate the effects of dimensionless radius of curvature and dimensionless area density on the ballistic limit velocity. On this basis, we investigated the influence of composite layup angle on the ballistic impact resistance.

5. We used machine learning methods, which was trained by a database from numerical simulations with various fiber contents, to investigate the relationship between microstructure and ballistic impact resistance of FRCs. The effects of different algorithms and parameters on the prediction model were also investigated, suggesting a suitable model for predicting the ballistic impact resistance of FRC.