Abstract

This paper addresses the influence of material and fabrication uncertainties on natural frequencies of curvilinear fiber laminate. Material properties and fiber orientation are considered uncertain. The sensitivity of input parameters to natural frequencies is obtained using high-dimensional model representation tool, and the results are compared with Monte Carlo simulation. This sensitivity pattern is established through various parametric studies. Highly sensitive parameters are identified and analyzed. Further stochastic analysis of natural frequencies of curvilinear fiber laminate is performed using polynomial neural network. The accuracy and efficiency of polynomial neural network are compared and conclusions are drawn.

摘要

本文讨论了材料和制造不确定性对曲线纤维层压板的固有频率的影响。材料特性和纤维取向被认为是不确定的。使用高维模型表示工具获得输入参数对固有频率的敏感性，并将结果与​​蒙特卡罗模拟进行比较。这种敏感性模式是通过各种参数研究建立的。识别和分析高度敏感的参数。使用多项式神经网络对曲线纤维层压板的固有频率进行进一步随机分析。对多项式神经网络的准确性和效率进行了比较并得出结论。