空间结构振动响应递归图构造方法研究

RESEARCH ON THE CONSTRUCTION METHOD OF RECURSIVE GRAPH OF VIBRATION RESPONSE OF SPATIAL STRUCTURE

文中以大跨空间结构为研究对象,为引入在数据特征提取方面具有独特优势的深度学习理论,提出一种新的基于振动信号的大跨空间结构分步损伤识别方法。方法利用递归图处理结构振动响应获取具有非线性特征的二维图像,通过图像中的非线性特征反映大跨空间结构不同损伤工况。结果表明使用这种方法处理得到的图像可以反应大跨度空间结构的损伤工况,可以为结构不同工况下的分类识别问题提供强有力的支撑。

Taking a large-span spatial structure as the research object,a new step-by-step damage recognition method for large-span spatial structure based on vibration signal is proposed with a deep learning theory with unique advantages in data feature extraction.The method employs recursive graphs to process vibration response of the structure to obtain a two-dimensional image with nonlinear features,which reflects the different damage conditions of the large-span spatial structure through the nonlinear features in the image.The results show that the images obtained by this method reflect the damage conditions of large-span spatial structures,providing strong supports for the classification and identification problems under different working conditions of structures.