文献类型：期刊文献

英文题名：Particle filter for fatigue crack growth prediction using SH0 wave on-line monitoring

作者：Li, Zhiwen[1];Jia, Jiuhong[1,2];Wang, Mingyuan[1];Gu, Mengqi[1];Tu, Shandong[1]

机构：[1]East China Univ Sci & Technol, Key Lab Pressure Syst & Safety, Minist Educ, Shanghai, Peoples R China;[2]East China Univ Sci & Technol, Sch Mech Engn, Meilong Rd 130, Shanghai 200237, Peoples R China

年份：2024

卷号：142

外文期刊名：ULTRASONICS

收录：;EI(收录号：20242316200894);WOS:【SCI-EXPANDED(收录号:WOS:001249765300001)】；

基金：This work was supported by the National Natural Science Foundation

语种：英文

外文关键词：SH0 wave monitoring; Particle filter; Damage index; Hydrogenation reactor; Fatigue crack growth prediction

摘要：Fatigue crack is one of the main failure modes of pressure vessels. Online monitoring and predicting methods of crack growth play an important role in the operation of important pressure vessel. The SH0 wave is nondispersive, and it is not disturbed by internal media of pressure vessel and very sensitive to cracks, therefore it is suitable for fatigue crack growth monitoring. Moreover, fatigue crack growth in industry is affected by material properties, loads, which usually shows some uncertainty. And the particle filter (PF) is well suited to deal with prediction problems affected by uncertainty. Hence, the prediction method of crack growth based on SH0 wave monitoring and PF is proposed (short for SH0-PF). The basic theory of crack monitoring method using SH0 wave is introduced, and the signal feature extraction using the damage index is studied. The state equation characterizing the fatigue crack growth is established by Paris model, and the observation equation is established based on the normalized correlation moment damage index according to monitoring signal using SH0 wave. The prediction reliability of the fatigue crack growth applying SH0-PF is verified by experiment with the single edge notched specimen. The experimental results indicate that the prediction accuracy of SH0-PF is better than that of the traditional Paris model.