Affiliations: [a] College of Electrical Engineering, Henan University of Science and Technology, Luoyang, China | [b] Power Electronics Device and System Engineering Laboratory of Henan, Henan University of Science and Technology, Luoyang, China
Correspondence:
[*]
Corresponding authors: Xi Li, College of Electrical Engineering, Henan University of Science and Technology, Luoyang 471023, China; Power Electronics Device and System Engineering Laboratory of Henan, Henan University of Science and Technology, Luoyang 471023, China. E-mail: [email protected]. Juwei Zhang, College of Electrical Engineering, Henan University of Science and Technology, Luoyang 471023, China; Power Electronics Device and System Engineering Laboratory of Henan, Henan University of Science and Technology, Luoyang 471023, China. Tel.: +86 0379 6562 7826; E-mail: [email protected]
Abstract: In order to solve the problems including poor signal denoising effect, low recognition rate, and poor real-time performance in wire rope magnetic flux leakage (MFL) testing, this paper proposes a new algorithm combining kernel extreme learning machine (KELM) with compressed sensing wavelet (CSW). Firstly, we consider a new mechanism and regularized orthogonal matching pursuit (ROMP) into CSW, and combine double-density wavelet transform (DD-DWT) to improve the result of wire rope signal noise reduction; Then, an effective normalization method is developed to improve the accuracy of classification. Finally, the detection accuracy and efficiency in wire rope quantitative identification are ameliorated through KELM. The effectiveness and novelties of the proposed algorithms are verified by the experimental platform based on unsaturated magnetic excitation non-destructive testing (NDT) device.
