GFRP管力学长期性能快速预测的回归优化方法  

Regression optimization method for rapid prediction of long-term property of GFRP pipes

作　　者：梁娜[1,2] 朱四荣[1] 陈建中[1] 方玺[1] 

机构地区：[1]武汉理工大学理学院,湖北武汉430070 [2]湖北科技学院数学与统计学院,湖北咸宁437100

出　　处：《华中科技大学学报（自然科学版）》2016年第8期128-132,共5页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：国家自然科学基金资助项目(11102142);湖北省教育厅资助项目(B2015083)

摘　　要：针对玻璃纤维增强塑料(GFRP)管道标准预测寿命50a力学性能值中,存在试样失效时间长达1×104h以上、数据量多达18个以上等问题,提出一种快速预测GFRP力学长期性能的新方法,基于回归优化理论,推导出短时预测方法及其改进方法.对GFRP管道应变腐蚀实验美国标准ASTM D—3681—01中算例数据,运用所提出方法和标准方法,分别建立双对数回归模型,预测GFRP管50a初始应变值,结果表明:与标准方法相比,仅含9个数据量的5 000h方法预测值的相对误差为5.63%,且含有10个数据量的1 000h改进方法3的预测值相对误差仅为0.55%,验证了所提出方法的有效性.For the long test duration up to 1×104 h and more than 18 data points in solving the prediction of long-term property of glass-fiber-reinforced plastic（GFRP）pipes under a condition of the current normative standard,a new method for rapid prediction of long-term property of GFRP pipes was proposed.The shorter time method and its improved method,which were based on regression optimization theory,were derived.With example data of the American standard ASTM D—3681—01,the data regression analysis of the proposed method and the standard method were presented.The results show that,compared to the standard method,the estimation error when using 5 000 hmethod with only 9data points is 5.63%,and the estimation error using 1 000 himproved method 3with 10 data points is only 0.55%.So that the validity of the proposed method has been clarified.

关 键 词：玻璃纤维增强塑料管道 力学长期性能 快速预测 双对数回归模型 回归优化 

分 类 号：TP13[自动化与计算机技术—控制理论与控制工程]