文献类型：期刊文献

中文题名：基于应变模态差和神经网络的管道损伤识别

英文题名：Damage Identification in Straight Pipeline Using Strain Modal Difference and Neural Network

作者：周邵萍[1];郝占峰[1];韩红飞[1];张佳程[1];章兰珠[1]

机构：[1]华东理工大学机械与动力工程学院,上海200237

年份：2015

卷号：35

期号：2

起止页码：334

中文期刊名：振动．测试与诊断

外文期刊名：Journal of Vibration,Measurement & Diagnosis

收录：CSTPCD;;EI(收录号：20152100880233);Scopus;北大核心:【北大核心2014】；CSCD:【CSCD2015_2016】；

基金：国家自然科学基金资助项目(51175178;51275171)

语种：中文

中文关键词：应变模态差;损伤识别;神经网络;管道

外文关键词：strain modal difference;;damage identification;;neural network;;pipeline

摘要：应变模态差对结构微小损伤具有很高的敏感性且对结构损伤处具有较高的定位识别率,故在工程实际中可以利用其对管道进行损伤识别。然而,应变模态差只能定性地反映结构的损伤程度,并不能直接量化损伤结构的损伤程度,故采用神经网络和应变模态差相结合的方法对损伤管道进行损伤位置和损伤程度的识别。利用有限元分析软件ANSYS进行模态分析提取管道的应变模态参数,并把管道损伤前后的应变模态差作为神经网络的输入参数,以损伤位置和损伤程度作为神经网络的输出参数,对损伤管道分别进行单损伤和双损伤的损伤定位和程度识别。研究结果表明,利用应变模态差和神经网络相结合的方法能够准确识别出管道的损伤位置以及损伤程度。
Strain modal difference is used to identify the location of damage in pipelines,due to its precise recognition in engineering practice and its sensitivity to small structural damage.However,this method fails to quantify the structure′s degree of damage,so a new method that combines the neural network with the strain modal differenceis presented to identify the degree of pipeline damage.It takes the strain modal difference obtained through the finite element as the input parameter and the damage location and degreeas the output parameter of the network to identify the location and degree of both single and double damage.The simulation results of damage identification in the pipeline show that this method can not only determine the location of pipeline damage,but also accuratelyquantify the degree of damage.