摘要
核电站高压蒸汽管道存在诸多问题,对他进行结构健康监测至关重要。文中以某核电站高压蒸汽管道为研究对象,建立缩比模型,结合有限元模型更新方法和BP神经网络进行损伤识别。通过管道模态采集试验获取结构特征参数,更新有限元模型,使他与实际管道结构在动态响应上相似。利用BP神经网络对损伤工况进行识别,选取合适的参数进行训练。结果表明训练后的神经网络在损伤识别中具有高准确率,证明了该方法的可靠性,为实际管道结构的损伤识别积累经验。
In nuclear power plants,high-pressure steam pipelines are subjected to many problems,so it is very important to monitor their structural health.In this paper,the scale model of a high-pressure steam pipeline is established for damage identification by the finite element model updating method and BP neural network.The characteristic parameters of the structure are obtained through the modal acquisition test of the pipeline,and the finite element model is updated to make it similar to the actual pipeline structure in dynamic response.BP neural network is used to identify the damage condition,and appropriate parameters are selected for training.The results show that the trained neural network has a high accuracy in damage identification.The reliability of the method is verified,providing the reference for the actual damage identification of pipeline structures.
作者
崔晓明
贾浩文
魏宗远
CUI Xiaoming;JIA Haowen;WEI Zongyuan(Heilongjiang Polytechnic of Architecture,Harbin 150001,China;Harbin Engineering University,Harbin 150001,China)
出处
《低温建筑技术》
2024年第6期6-10,共5页
Low Temperature Architecture Technology
基金
国家自然科学基金项目“基于主被动声发射融合的大跨度桥梁钢箱梁疲劳裂纹智能监测与诊断”(52278297)。
关键词
损伤识别
神经网络
管道结构
固有频率
模态振型
高压管道
damage identification
neural network
pipeline structure
natural frequency
mode shape
high-pressure pipeline
