西安建筑科技大学学报(自然科学版)

为了解决应用金属磁记忆技术判定结构初期损伤状态的问题,从建筑钢板件的三点受弯试验入手,通过研究试验结果的磁记忆曲线规律,并提取四维磁记忆特征向量,结合支持向量机的方法建立了一个判定焊缝损伤状态的算法.结果表明:磁记忆检测技术不仅适用于建筑钢结构,而且可以利用支持向量机小样本的优势将焊缝损伤在应力集中、隐性损伤、显性损伤三种类别中区分出来,在一定程度上解决了如何识别钢结构隐性损伤的难题,并为磁记忆评判损伤量化提供了一种方法.

It is difficult for the metal magnetic memory(MMM)technology to test the initial weld damage state and category in the detection of building steel structure.To solve the problem, the research started from the three point bending test specimens of building steel structure. By analyzing the regular pattern of the magnetic memory curve and extracting the four-dimensional magnetic memory parameter, an algorithm of determining weld damage categories combined with support vector machine(SVM)method is established. Results show that the metal magnetic memory technology is not only suitable for building steel structure, but also capable of combining with the advantage of support vector machine in small sample to distinguish the category of weld damage from stress concentration, implicit damage and visible damage. The research solves the decision problem of steel structure implicit damage to a certain extent, and provides a method for quantitative study of magnetic memory.

引言
1 支持向量机原理
2 试验方案
3 试验结果
4 结论

图1 核函数转化原理图<br/>Fig.1 Schematic diagram of nuclear function transformation

图1 核函数转化原理图
Fig.1 Schematic diagram of nuclear function transformation

图2 三点弯曲试验的试件尺寸(单位:mm)<br/>Fig.2 Specimen size of three point bending test(Unit:mm)

图2 三点弯曲试验的试件尺寸(单位:mm)
Fig.2 Specimen size of three point bending test(Unit:mm)

图3 试验设备
Fig.3 Test equipment

表1 试件的损伤状态评估情况<br/>Tab.1 Defect assessment status of test piece

表1 试件的损伤状态评估情况
Tab.1 Defect assessment status of test piece

图4 法向分量HP(y)的分布规律<br/>Fig.4 Distribution law of normal component HP(y)

图4 法向分量HP(y)的分布规律
Fig.4 Distribution law of normal component HP(y)

图5 法向分量梯度K的分布规律<br/>Fig.5 Distribution law of normal component gradient K

图5 法向分量梯度K的分布规律
Fig.5 Distribution law of normal component gradient K

表2 无焊接缺陷试件的评估结果<br/>Tab.2 Evaluation results on samples without welding damage

表2 无焊接缺陷试件的评估结果
Tab.2 Evaluation results on samples without welding damage

表3 有焊接缺陷试件的评估结果<br/>Tab.3 Evaluation results of samples with weld damage

表3 有焊接缺陷试件的评估结果
Tab.3 Evaluation results of samples with weld damage

表4 三点弯曲试验无焊接缺陷试件的原始数据<br/>Tab.4 Raw data of samples free of weld damage in three point bending test

表4 三点弯曲试验无焊接缺陷试件的原始数据
Tab.4 Raw data of samples free of weld damage in three point bending test

表5 三点弯曲试验有焊接缺陷试件的原始数据<br/>Tab.5 Raw data of welding damage samples in three point bending test

表5 三点弯曲试验有焊接缺陷试件的原始数据
Tab.5 Raw data of welding damage samples in three point bending test

表6 损伤识别准确度<br/>Tab.6 Accuracy of defect identification

表6 损伤识别准确度
Tab.6 Accuracy of defect identification

[1] 王威, 曾发荣, 苏三庆,等. 基于磁记忆的受弯钢梁力磁效应试验[J]. 材料科学与工程学报, 2016,34(1): 109-114.
[2] 王威, 任广超, 苏三庆, 等. 钢梁弯曲应变与磁记忆效应的关系研究[J]. 西安建筑科技大学学报(自然科学版), 2017, 49(1):29-35.
[3] DUBOV A A.Development of a metal magnetic memory method[J]. Chemical and Petroleum Engineering, 2012,(47):837-839.
[4] DUBOV A A.Diagnostics of equipment and constructions strength with usage of Magnetic Memory Inspection[J]. Diagnostics.2011,35(6):19-29.
[5] DUBOV A A, DEMIN E A, MILYAEV A I. The experience of gas pipeline stress-strain state control with usage of the metal magnetic memory and stress control means[J]. Welding in the World,2002,46(9):29-33.
[6] 宗周红, 钟儒勉, 郑沛娟, 等. 基于健康监测的桥梁结构损伤预后和安全预后研究进展及挑战[J]. 中国公路学报, 2014, 27(12):46-57.
[7] 易方, 李著信, 吕宏庆, 等. 基于模糊核支持向量机的管道磁记忆检测缺陷识别[J]. 石油学报, 2010, 31(5):863-866.
[8] BURGES C J C. Atutoriol on support vector machines for pattern recognition[J]. Data Mining and Knowledge Discovery,1998,2(2):121-167.
[9] 王建国, 张文兴. 支持向量机建模及其智能优化[M].北京:清华大学出版社,2015.
[10]邸新杰, 李午申, 严春妍, 等. 基于金属磁记忆的宏观焊接裂纹识别方法[J]. 中国机械工程, 2007(12):1475-1478.
[11]HSU C W, CHANG C C, LIN C J. A practical guide to support vector classification [R]. Taipei:Department of Computer Science National Taiwan University, 2003.

备注

引言

1 支持向量机原理

2 试验方案

3 试验结果

4 结论

期刊信息