预测碳纤维薄板增强RC梁抗弯疲劳寿命的AFN法.pdfVIP

第4卷 第6期 铁道科学与工程学报 Vo1．4 No．6 2007年 l2月 JOURNAL OF RAII WAY SCIENCE AND ENGINEERING Dec．200r7 预测碳纤维薄板增强RC梁抗弯疲劳寿命的AFN法 彭可可，邓 军，黄培彦 (华南理工大学交通学院，广东广州510640) 摘 要：在碳纤维薄板(CFL)增强Rc梁中，加固后构件的疲劳寿命受到被加固结构本身性能、加固材料性能以及荷载水平 等因素的影响，加固效果具有明显的非线性，加固后疲劳寿命难以显式表达。本文采用AFN法预测加固梁的疲劳寿命：用 层次分析：(Analytic hierarchy process)对影响疲劳寿命的主要因素进行分析；用模糊聚类法(Fuzzy clustering)确定训练样本； 运用神经网络方法(Neural network)，建立疲劳寿命隐函数关系式，从而预测CFL加固梁抗弯疲劳寿命。以疲劳试验数据为 样本，用层次分析法选取最大荷载、CFL的应力水平、初始挠度、循环次数为200时跨中挠度的试验数据作为疲劳寿命隐函 数的随机变量输入单元，以CFL加固梁的抗弯疲劳寿命作为输出单元。通过预测值与试验值的对比分析，验证了AFN法预 测CFL增强RC梁抗弯疲劳寿命的合理性。 关键词：AFN法；疲劳寿命；碳纤维薄板；RC梁；神经网络；层次分析( )；模糊聚类 中图分类号：0346．2；U441 文献标识码：A 文章编号：1672—7029(2007)06—0030—05 The AFN method for predicating bending fatigue life of RC beams strengthened with CFL PENG Ke—ke，DENG Jun，HUANG Pei—yan (College ofTraffic and Communications，South China University ofTechnology，Guangzhou 510640，China) Abstract：The fatigue life of the reinforced concrete(RC)beams strengthened with carbon fiber laminate(CFL)was influenced by the properties of the RC beam，the strengthening materials and the loading conditions，etc．Therefore， the strengthening effect has high nonlinearity．In order to predict the fatigue life of the strengthened beams，the AFN (analytic hierarchy process—fuzzy clustering—neural network)method Was introduced．Main factors influencing the fatigue life were analyzed by analytic hierarchy process(AHP)and the training sample Was selected by fuzzy clustering method．The training samples were chosen from the test results obtained by the authors’research group．The neural network model included four input elements，which were the maximum load，the stress range in CFL，the initial de— flection and the deflection w