为了有效评价测量响应中不确定性对结构参量识别结果的影响,提出一种基于λ概率密度函数(Probability distribution function,PDF)和一次二阶矩的不确定性计算反求方法。采用二次衍生λ-PDF对待识不确定性参量的PDF进行建模。内层通过对参量呈λ-PDF的功能函数采用一次二阶矩法进行正问题求解,得到计算响应的概率分布;外层通过最小化测量响应与计算响应之间的概率分布特征量将不确定性反问题转化为确定性的最优化问题,并用隔代映射遗传算法识别未知参量λ-PDF的参数。本方法不仅有效地实现了结构未知参量PDF的估计,而且与传统基于抽样的统计方法相比,计算效率较高。数值算例和工程应用验证了本方法的可行性和有效性。
An experimental procedure is conducted to investigate the mechanism of the non-monotonic characteristic between curing temperature and mechanical behaviors of the Kevlar/epoxy composite in macro and micro levels. Different specimens are fabricated at four different curing temperatures and tested with ±45° off-axis tensile loading on a universal test machine coupled with digital image correlation(DIC). Moreover, the environmental scanning electron microscope(SEM) was used to obtain the micrographs and reveal their mechanism. The tested results show that the tensile mechanical behaviors are sensitive to the curing temperature and the relationship is non-monotonic. Also, as the temperature increases, the thicknesses of the specimens are significantly enlarged. By analyzing the SEM micrographs of the matrix grooves in the damage zone and DIC strain contours, it is concluded that the non-monotonic relationship is dominated by the properties of the Kevlar/epoxy interfaces and deformation of the distorted fibers.
