Abstract:

To accurately detect wire rope damage signals and extract effective features, an adaptive fast Walsh-Hadamard transform (FWHT) method was proposed for noise reduction and feature extraction of wire rope damage signals. The noisy magnetic flux leakage signals were transformed into the sequency domain. An adaptive particle swarm optimization (APSO) algorithm was employed to search for the optimal number of retained transform coefficients. With a reference signal, the cross-correlation coefficient was used as the fitness function. In the absence of a reference signal, a composite index consisting of the peak-to-background ratio, local energy concentration, and residual energy ratio was adopted as the fitness function. Adaptive reconstruction of damage signals under strong-noise conditions was thus achieved. Experimental results show that, under strong-noise conditions with SNRs of -10 and -20 dB, the proposed adaptive FWHT method provides better restoration of damage peak values and preservation of peak amplitude information for broken wire defects than the Sym4 and db8 wavelet denoising methods.

Key words: wire rope, magnetic flux leakage signal, fast Walsh-Hadamard transform (FWHT), adaptive particle swarm optimization (APSO), signal denoising