Abstract Complex structural components made from 7xxx series alloys are usually manufactured through hot stamping due to their low ductility at room temperature. With the help of a custom-made induction heating system, experiments are performed on aluminum alloy 7075 in its W-temper. The comprehensive experimental program is executed comprising more than 100 specimens to characterize the plasticity and fracture response at temperatures ranging from 180 to 480 °C and strain rates ranging from 0.001 to 2/s. Stress states ranging from simple shear to biaxial tension are obtained through the use of planar shear specimens and tensile specimens with central holes and various notches. Based on the experimental results, an attempt is made to calibrate an enhanced Johnson-Cook type of constitutive model before employing an isotropic machine learning based constitutive model for hybrid experimental-numerical analysis of the fracture experiments. The extracted loading paths to fracture reveal a negative strain rate effect and a non-monotonic effect of temperature on the ductility of AA7075-W. An isotropic neural network based fracture initiation model is proposed, trained and validated to describe the onset of fracture across the range of stress states, strain rates and temperatures considered.

中文翻译：

---

7075 铝合金的应变率和温度相关断裂：实验和神经网络建模

摘要 7xxx 系列合金制成的复杂结构件，由于其室温延展性低，通常采用热冲压制造。在定制的感应加热系统的帮助下，对铝合金 7075 的 W 回火进行了实验。执行的综合实验程序包括 100 多个试样，以表征温度范围为 180 至 480 °C 和应变速率范围为 0.001 至 2/s 下的塑性和断裂响应。通过使用具有中心孔和各种缺口的平面剪切试样和拉伸试样，可以获得从简单剪切到双轴拉伸的应力状态。根据实验结果，在采用基于各向同性机器学习的本构模型对断裂实验进行混合实验-数值分析之前，尝试校准增强型 Johnson-Cook 类型的本构模型。提取的断裂加载路径显示负应变率效应和温度对 AA7075-W 延展性的非单调效应。提出、训练和验证基于各向同性神经网络的断裂起始模型，以描述在所考虑的应力状态、应变率和温度范围内断裂的发生。