基于XFEM与BP神经网络的裂纹智能识别.pdfVIP

第39卷第6期 应 用 力 学 学 报 Vol.39No.6 2022年12月 Chinese Journal of Applied Mechanics Dec.2022 文章编号： 2)06-1158-10 基于XFEM与B 神经网络的裂纹智能识别 毛晓敏，张慧华，纪晓磊，韩尚宇 （南昌航空大学土木建筑学院，330063南昌） 摘要：数值技术与智能算法的飞速发展为结构内部缺陷的识别提供了新途径。本研究建立了扩 展有限元法（XFEM ）与误差反向传播多层前馈（BP）神经网络相结合的裂纹反演分析模型。模型通 过XFEM正向分析获得的测点位移数据训练BP神经网络，在此基础上利用该网络进行裂纹反向识 别。通过2个典型算例验证了模型的可行性和精度，结果表明本研究的方法能够准确反演裂纹的 几何信息，与此同时还探讨了测点布置方式及输入数据噪声等因素对识别精度的影响。 关键词：扩展有限元法；BP神经网络；裂纹；反演分析 中图分类号：034 文献标志码：A D0I:10.11776/j.issn.1000-4939.2022.06.017 Intelligent crack identification based on XFEM and BP neural network MAO Xiaomin,ZHANG Huihua,JI Xiaolei,HAN Shangyu (School of Civil Engineering and Architecture, Nanchang Hangkong University,330063 Nanchang, China) Abstract : The rapid development of numerical technology and intelligent algorithm provides a new way to identify the internal defects of structures. In this paper , an inverse analysis model for crack detection is es- tablished by combining extended finite element method feedforward (BP) neural network. The BP neural network is trained by the displacement data obtained from the forward analysis of XFEM. On this basis, the network is used for the inverse identification of cracks. The feasibility and accuracy of the model are verified by two typical examples. The results show that the proposed method can accurately retrieve the geometric information of cracks. At the same time,the influence of the layout of measuring points and the input data noise on the identification accuracy is also discussed. Key words:extended finite element method; BP neural network; crack; inverse analy