论文部分内容阅读
损伤识别方法是结构健康监测系统的重要组成部分。基于广义回归神经网络(GRNN)模型,建立了结构损伤识别的两步法,构造了用于损伤定位和损伤定量的不同损伤识别组合损伤指标,并引入模态应变能系数选择节点,最后,结合典型桁架结构进行了损伤识别数值模拟研究。结果表明,即使在只获得低阶频率和少量节点一阶振型数据且含有噪声的情况下,采用构造的组合参数,GRNN神经网络对损伤位置及损伤程度识别都取得了比较理想的识别效果。
Damage identification is an important part of the structural health monitoring system. Based on generalized regression neural network (GRNN) model, a two-step method of structural damage identification is established, and the damage indexes of different damage identification combinations for damage localization and damage quantification are constructed, and modal strain energy coefficient selection nodes are introduced. Finally, A typical truss structure is identified by numerical simulation of damage identification. The results show that the GRNN neural network can achieve better identification of the damage location and degree of damage even with only low-order frequencies and a small number of first-order mode shapes and noises.