Accurate Detection of Weak Defect Signals in Wire Rope via Magnetic Flux Leakage Using Physically Guided Variable Symmetric Gaussian Wavelet

Accurate detection of weak wire rope (WR) defect signal is a significant challenge in magnetic flux leakage (MFL) testing. To address this, a novel variable symmetry Gaussian (VSG) wavelet basis is developed, and an adaptive feature extraction method is proposed. First, the sources of weak defect signals, including small-sized defects and attenuation caused by large liftoff, are analyzed, and the variable symmetry characteristics of the damage signals on the helical structure of the WR at different detection angles are found and explained. Second, a novel VSG wavelet is designed based on the physical helical structure of the WR, which can flexibly adjust the direction and degree of variable symmetry to enhance the similarity with the defective signals while reducing the similarity with the noise components. Finally, an interquartile range combined with the upper tail range (IQ-UT) feature is proposed to adaptively select convolutional features for improved weak signal identification. The proposed method utilizes signal variable symmetry characteristics to construct the VSG wavelet basis for adaptive weak defect signal detection. Comparative experiments with various methods demonstrate the superior accuracy of the proposed method.