基于神经网络的平面钢桁架结构损伤识别研究

利用神经网络的鲁棒性、容错性和泛化能力,建立了3个不同的神经网络对平面钢桁架结构进行了损伤定位和定量的评估。首先用PNN神经网络诊断出损伤杆件所在的子结构;并用RBF神经网络进一步诊断出损伤杆件的具体位置;进而确定出损伤杆件的损伤程度。数值仿真表明,该方法用于平面钢桁架结构的损伤识别是可行的。

Research on Damage Identification of Plane Steel Truss Based on Neural Network

Utilizing robustness,fault toleration and generalization ability of neural network,three different neural networks were established to evaluate the localization and quantification of the plane steel truss damage.firstly,PNN neural network was employed to locate the damaged substructure and RBF neural network was employed to further locate the specific damaged bar.Finally,RBF neural network was employed to quantify the damage extent.Numerical simulation shows that this method is effective.