Research on 3D MFL testing of wire rope based on empirical wavelet transform and SRCNN

Abstract

Magnetic flux leakage (MFL) testing is one of the most effective methods in nondestructive testing of wire rope. However, traditional MFL testing devices have problems such as low recognition rate, single detection dimension and fuzzy magnetic leakage image. Based on the non-saturated magnetic excitation 3D MFL testing device, this paper proposes a wavelet denoising method based on empirical wavelet transform (EWT) to denoise the collected 3D MFL signal. After noise reduction, the Signal-to-Noise Ratio (SNR) and Root Mean Squared Error (RMSE) of the three dimensions have improved. Color imaging technology is used to fuse defect grayscale images into color images, and Super-Resolution Convolutional Neural Network (SRCNN) is applied to MFL images of broken wires. After SRCNN reconstruction, the resolution of defect color images is improved. The color moment feature of the defect color image is extracted as the input of the Elman neural network to quantitatively identify broken wires. Experimental results show that the noise reduction algorithm can effectively suppress the noise in three dimensions, and the broken wires recognition rate after reconstruction has been significantly improved, which verifies the effectiveness of SRCNN in wire rope MFL images.