Defect reconstruction from MFL signals using improved genetic local search algorithm

Author

Han, Wenhua ; Que, Peiwen

Author_Institution

Inst. of Autom. Detection, Shanghai Jiao Tong Univ.

fYear

2005

fDate

14-17 Dec. 2005

Firstpage

1438

Lastpage

1443

Abstract

This paper presents an improved GLSA (IGLSA) by incorporating the simulated annealing technique into the perturbation process of the genetic local search (GLSA), and proposes an IGLSA-based inverse algorithm for 2-D defect reconstruction from the magnetic flux leakage (MFL) signals. In the algorithm, radial-basis function neural network (RBFNN) is utilized as forward model, and the IGLSA is used to solve the optimization problem in the inverse problem. Experiments are presented to show the performance of the IGLSA-based inverse algorithm and to compare it with the canonical-genetic-algorithm based (CGA-based) inverse algorithm and the GLSA-based inverse algorithm, respectively. The results demonstrate that IGLSA-based inverse algorithm is more accurate and is robust to the noise

Keywords

electrical engineering computing; genetic algorithms; magnetic flux; magnetic leakage; radial basis function networks; signal reconstruction; simulated annealing; 2D defect reconstruction; RBFNN; canonical-genetic-algorithm based inverse algorithm; improved genetic local search algorithm; magnetic flux leakage; optimization problem; perturbation process; radial-basis function neural network; simulated annealing technique; Genetic algorithms; Inverse problems; Iterative methods; Magnetic flux leakage; Neural networks; Noise robustness; Predictive models; Shape measurement; Signal processing; Simulated annealing;

fLanguage

English

Publisher

ieee

Conference_Titel

Industrial Technology, 2005. ICIT 2005. IEEE International Conference on

Conference_Location

Hong Kong

Print_ISBN

0-7803-9484-4

Type

conf

DOI

10.1109/ICIT.2005.1600861

Filename

1600861