论文部分内容阅读
神经网络已越来越多地被用于基于振动的结构损伤识别 ,而如何选择输入参数还是一个值得研究的问题。本文提出了一种由固有频率与少数点的模态分量合成的组合参数 ,作为神经网络的输入向量 ,以克服单独使用某种参数的缺陷。通过一个六层框架的数值模拟和一个两层框架的实验验证 ,表明本文提出的组合参数对框架结构的连接损伤识别是实用可行的
Neural networks have increasingly been used to identify structural damage based on vibration, and how to choose input parameters is still a problem worth studying. In this paper, a combination of natural frequencies and modal components of a few points is proposed as an input vector to the neural network to overcome the disadvantage of using a certain parameter alone. Through a numerical simulation of a six-story frame and a two-level frame experiment, it is proved that the combination parameters proposed in this paper are practical and feasible for the identification of the connection damage of the frame structure