基于XFEM的大体积结构波动传播规律及裂纹反演方法  

Wave Propagation and Crack Detection Method for Massive Structures Based on XFEM

作　　者：卢皓卓 江守燕[1] LU Haozhuo;JIANG Shouyan(Department of Engineering Mechanics,Hohai University,Nanjing 211100,China)

机构地区：[1]河海大学工程力学系,南京211100

出　　处：《三峡大学学报（自然科学版）》2024年第1期23-29,共7页Journal of China Three Gorges University：Natural Sciences

基　　金：国家自然科学基金项目(52279130);中国地球物理学会工程物探检测重点实验室开放研究基金项目(CJ2021GE06)。

摘　　要：大体积混凝土结构被广泛应用于土木、水利等领域的重大工程中,而混凝土抗拉强度低的力学特性决定了其易产生裂纹,因此,发展高效的检测方法,识别大体积混凝土结构中的裂纹信息十分必要.论文提出了一种新的方法,通过提取响应信号频谱中特定频率的幅值特征,基于BP人工神经网络建立幅值特征与裂纹信息间的映射关系,从而有效识别出裂纹信息.首先采用扩展有限元法(eXtended Finite Element Methods, XFEM)和人工吸收边界模型,分别模拟了单裂纹和双裂纹情形下,大量不同裂纹信息下特定位置传感器的响应,分析其频谱曲线并提取特征,建立频谱特征—裂尖位置数据集,以训练人工神经网络,测试集的反演效果显示,该方法具有较好的准确度,可有效识别出裂纹信息.Massive concrete structures are widely used in major engineering projects such as civil engineering and water conservancy.The low tensile strength of concrete determines its susceptibility to cracking.Therefore,it is necessary to develop efficient detection methods to identify crack information in massive concrete structures.The paper proposes a new method to extract amplitude features of specific frequencies in the response signal spectrum,and establish a mapping relationship between amplitude features and crack information based on BP artificial neural network,thereby effectively identifying crack information.Firstly,the eXtended Finite Element Methods(XFEM)and artificial absorbing boundary layer models were used to simulate the response of specific position sensors under a large amount of different crack information,in the case of single and double cracks,respectively.Then,the spectral curves were analyzed and features were extracted.Finally,a dataset of spectral features and crack tip positions was established to train the artificial neural network.The inversion results of the test set showed that this method has high accuracy,which can effectively identify crack information.

关 键 词：大体积结构 裂纹反演 频域特征 神经网络 扩展有限元法 吸收边界层 

分 类 号：TV331[水利工程—水工结构工程]