Advanced studies on eddy current testing using computational intelligence related to electromagnetic inverse problems
Grant-in-Aid for Scientific Research (C)
|Allocation Type||Single-year Grants |
|Research Institution||KOBE UNIVERSITY (1999-2000)|
Osaka Institute of Technology (1998)
KOJIMA Fumio Kobe University Graduate School of Science and Technology, Professor, 自然科学研究科, 教授 (70234763)
KUBOTA Naoyuki Fukui University Faculty of Engineering, Associate Professor, 工学部, 助教授 (30298799)
友枝 謙二 大阪工業大学, 工学部, 教授 (60033916)
高木 敏行 東北大学, 流体科学研究所, 教授 (20197065)
|Project Period (FY)
1998 – 2000
Completed (Fiscal Year 2000)
|Budget Amount *help
¥3,000,000 (Direct Cost: ¥3,000,000)
Fiscal Year 2000: ¥800,000 (Direct Cost: ¥800,000)
Fiscal Year 1999: ¥800,000 (Direct Cost: ¥800,000)
Fiscal Year 1998: ¥1,400,000 (Direct Cost: ¥1,400,000)
|Keywords||Nuclear engineering / Nondestructive testing / Material evaluation / Inverse problem / Soft computing / System integrity / System analysis / Modeling / 逆問題解析 / 計算知能工学 / 電磁工学 / システム構成学|
Our mission is to develop feasible computational methods for quantitative nondestructive evaluation related to eddy current testing with the background knowledge of computational intelligence. During the period of the research, our efforts were directed to the developments of the following inverse solvers.
(1) Shape recovering algorithm using the GP-based fuzzy inference system
A quantitative nondestructive evaluation in eddy current testing was proposed by using genetic programming (GP) and fuzzy inference system. GP is applied to extract and select effective features from a probe impedance trajectory. With the use of the extracted features, a fuzzy inference system could detect presence, position, and size of a defect of test sample.
(2) Fast computational inverse solver using the multivariate splines
In this method, the model output was directly reconstructed by parameter space with data sets of the forward solutions. Hence the sensitivities of the output least square problem could be e
valuated very fast. As a result, the tremendous amounts of computational savings have been achieved using the proposed scheme.
(3) Evolutionary computation and its application to QNDE
A coevolutionary algorithm (CEA) was effectively used for a crack shape identification problem where the number of shapes is unknown. A CEA including two populations of candidate shapes and of candidate combination was proposed. Results were demonstrated that the proposed method makes it possible to obtain the number of cracks as well as to identify their shapes.
(4) Shape identification using the SQUID based NDE system
A quantitative nondestructive evaluation of using superconducting quantum interference devices (SQUIDs) was developed for conducting materials with a depth-varying crack. A computational method based on the genetic algorithm was proposed for recovering internal defect profiles with SQUID data.
Consequently, all computations were successfully tested for the JSAEM benchmark problems, including the masked tube samples with natural cracks. Less
Report (4 results)
Research Products (73 results)