裂纹反演分析的NMM-Elman神经网络协同方法

Crack inverse analysis with the NMM-Elman neural network collaborative method

裂纹识别是结构健康监测的重要内容。本研究基于反演分析原理,将数值流形方法(NMM)与Elman神经网络相结合开展裂纹识别。NMM用于获取对应裂纹构型下测点的位移数据以供Elman神经网络的学习,在此基础上利用训练好的Elman网络进行了直线裂纹反演。通过2个典型算例证实了NMM-Elman协同方法的可行性和精度,与此同时分析了测点布置方式及输入数据噪声等因素对裂纹反演精度的影响。表明本研究的方法能够准确反演出单一及复杂裂纹的裂尖坐标。本研究的工作为复杂裂纹的高效准确识别提供了一种新的思路和方法。

Crack identification is an important issue in structural health monitoring.Based on the principle of inverse analysis,this paper combines the numerical manifold method(NMM)with the Elman neural network to carry out crack identification.To serve the learning of Elman neural network,the NMM is used to obtain the displacement data of measuring points under various crack configurations.On this basis,the trained Elman network is used for straight crack inversion.The feasibility and accuracy of NMM-Elman collaborative method are verified by two typical examples.At the same time,the effects of measuring point layout and input data noise on crack inversion accuracy are analyzed.The research shows that the method proposed in this paper can accurately reflect the crack tip coordinates of single and complex cracks.This work provides a new pathway for efficient and accurate detection of complex cracks.