基于模糊加权马尔科夫链的焊缝隐性损伤磁记忆特征参数定量预测

发布时间：2018-02-05 17:56

本文关键词： 隐性损伤磁记忆无偏灰色模糊隶属度马尔科夫链　出处：《机械工程学报》2017年12期 　论文类型：期刊论文

【摘要】：针对磁记忆技术在焊缝隐性损伤状态定量评价与预测中的难题,建立基于无偏灰色模糊加权马尔科夫链的焊缝隐性损伤状态磁记忆特征参数定量预测模型。以Q235焊接试件为试验材料,拾取焊缝纵向和横向磁记忆信号分布,与同步X射线检测结果进行对比得出:在宏观损伤时满足传统的磁记忆检测判据,即切向分量Hp(x)具有极大值,法向分量Hp(y)过零值;而在隐性损伤时并没有出现传统判据特征,说明Hp(x)与Hp(y)并不能准确地判断隐性损伤状态和程度。为此引入正交矢量合成梯度比Kr,可以较为敏感地反映隐性损伤状态,但敏感性同时带来局部振荡波动性,造成对隐性损伤后续发展状态的磁记忆定量预测困难。因此,在无偏灰色预测理论和马尔科夫链的基础上,结合模糊隶属函数建立模糊加权马尔科夫链磁记忆定量预测模型,验证结果表明:无偏灰色理论结合模糊加权马尔科夫链后的定量预测模型,其最大相对误差从38.492 5%降到5.046 4%,为实际工程中焊缝隐性损伤未来发展状态的定量预测与维修策略选择提供了新的思路。
[Abstract]:Aiming at the difficult problem of magnetic memory technology in quantitative evaluation and prediction of hidden damage state of weld seam. A quantitative prediction model of magnetic memory characteristic parameters of hidden damage state of weld seam based on unbiased grey fuzzy weighted Markov chain was established. The Q235 welding specimen was used as the test material. The longitudinal and transverse distribution of magnetic memory signals of weld seam is picked up and compared with the results of synchronous X-ray detection. It is concluded that the traditional criterion of magnetic memory detection is satisfied in macroscopic damage, that is, the tangential component HpPX) has the maximum value. Normal component Hpt0) zero crossing value; However, there is no traditional criterion feature in recessive damage, which shows that HpX) and HpCY) can not accurately judge the state and degree of recessive damage. Therefore, orthogonal vector synthesis gradient ratio Kr is introduced. It can reflect recessive damage state sensitively, but sensitivity brings local oscillation fluctuation at the same time, which makes it difficult to quantitatively predict the future development state of recessive damage by magnetic memory. On the basis of unbiased grey prediction theory and Markov chain, a fuzzy weighted Markov chain magnetic memory quantitative prediction model is established based on fuzzy membership function. The results show that the maximum relative error of the model is reduced from 38.492 5% to 5.046 4% with the combination of unbiased grey theory and fuzzy weighted Markov chain. It provides a new way of quantitative prediction and maintenance strategy selection for the future development state of hidden damage of weld seam in practical engineering.
【作者单位】：东北石油大学机械科学与工程学院;
【基金】：国家自然科学基金(11272084,11072056,11472076) 中石油科技创新基金(2015D-5006-0602) 东北石油大学研究生创新科研(YJSCX2016-024NEPU) 黑龙江省博士后科研启动基金(LBH-Q13035)资助项目
【分类号】：TG441.7

【相似文献】