Reconstruction of 3-D defect profiles from MFL signals using radial wavelet basis function neural network

Issue title: Proceedings from the 16th International Symposium on Applied Electromagnetics and Mechanics (ISEM 2013)

Guest editors: Xavier Maldague and Toshiyuki Takagi

Article type: Research Article

Authors: Chen, Junjie^a | Huang, Songling^{a; *} | Zhao, Wei^a | Wang, Shen^a

Affiliations: [a] State Key Laboratory of Power System and Department of Electrical Engineering, Tsinghua University, Beijing, China | Université Laval, Canada | Tohoku University, Japan

Correspondence: [*] Corresponding author: Songling Huang, State Key Laboratory of Power System and Department of Electrical Engineering, Tsinghua University, Beijing 100084, China. E-mail: huangsling@tsinghua.edu.cn

Abstract: This paper proposes an inversion procedure, based on radial wavelet basis function (RWBF) neural network, to reconstruct 3-D defect profiles from magnetic flux leakage (MFL) data. The architecture of the neural network, the adaptive training algorithm and the reconstruction process are presented. Defects reconstructed from both simulated and experimental MFL data, together with comparison with two other inversion methods, demonstrate the efficiency and accuracy of the proposed inversion procedure.

Keywords: MFL, defect reconstruction, RWBF neural network

DOI: 10.3233/JAE-141865

Journal: International Journal of Applied Electromagnetics and Mechanics, vol. 45, no. 1-4, pp. 465-471, 2014