| 研究課題/領域番号 |
20J11957
|
|---|
| 研究機関 | 東北大学 |
|---|
研究代表者 |
宋 海成 東北大学, 工学研究科, 特別研究員(DC2)
|
|---|
| 研究期間 (年度) |
2020-04-24 – 2022-03-31
|
| キーワード | Probability of detection / Inverse analysis / Reliability analysis / Local wall thinning / Sensor placement |
|---|
| 研究実績の概要 |
A model has been proposed to quantify the detection capability of low frequency electromagnetic monitoring method to local wall thinning by considering not only multiple dimensions of the defect but also sensor placement. Part of this study was presented at ICFD2020. In addition, an inverse algorithm has been developed to estimate the size of local wall thinning probabilistically with the help of multivariate probabilistic mode to quickly generate large training data. This study has been published as a paper on the journal of NDT&E International. A defect growth prediction model was adopted to assess the reliability of pipe by considering multi-variable POD model, and the research content was presented at COMPSAFE2020.
|
| 現在までの達成度 (区分) |
現在までの達成度 (区分)
1: 当初の計画以上に進展している
理由
Due to the pandemic of COVID-19, the research activities have been significantly affected, which makes it difficult to perform researches as usual. Despite these, during the fiscal year, more advanced models have been proposed to quantify the detection uncertainty and evaluation uncertainty of low frequency electromagnetic method to local wall thinning and their validity have been examined by experiments. Moreover, for the reliability analysis of pipe suffering wall thinning, current models have been reviewed. Two defect growth prediction models have been adapted to incorporate the quantifications of the uncertainties to predict pipe failure probability. All of the research processes can basically keep up with the research plan in the proposal.
|
| 今後の研究の推進方策 |
For quantifying the detection uncertainty, the proposed probability of detection model will be improved to consider more factors related to practical wall thinning, say, defect morphology, which have significant effect on signal response.
For quantifying the evaluation uncertainty, an inverse algorithm will be explored to estimated the size of wall thinning based on monitoring signals that are sparse and limited.
For assessing the reliability of pipe suffering wall thinning, a model will be developed to properly predict the failure probability of pipes by considering the output of the proposed models for quantifying the uncertainties of low frequency electromagnetic monitoring method. And its applicability to other nondestructive method will also be examined.
|
