Quantitative Nondestructive Testing of Broken Wires for Wire Rope Based on Multi-Image Fusion

Nondestructive testing (NDT) has been increasingly applied to wire rope defect detection. In this experiment, in order to obtain better classification effect of wire rope defects, a broken wire identification method that fuses the magnetic flux leakage (MFL) signal, thermal infrared signal and visible light signal is proposed. The MFL signal is processed by the sliding mean filtering and the MFL signal denoising algorithm based on Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) to remove the baseline and high frequency noise from the MFL signal. A infrared image defect extraction method based on wire rope strand structure is proposed. Morphological processing of visible images is performed to extract visible defect images and applying the gamma transform enhances them. Features are extracted from the three kinds of defect images and feature fusion is performed. A Self-Organizing Feature Map (SOFM) network is employed for quantitative identification of broken wires. Experiment results show that the fusion of three kinds of defect images features can further improve the recognition effect of broken wires.