Structural damage monitoring for metallic panels based on acoustic emission and adaptive improvement variational mode decomposition–wavelet packet transform

Abstract

The metallic panel acoustic emission signal with strong non-stationary properties is normally composed of multiple components (e.g. impulses, background noise, and other external signal), where impulses relevant to metallic panel are easily contaminated by background noise and other external signal, making it difficult to excavate the inherent acoustic emission signal features. To address this issue and achieve the damage monitoring of metallic panels based on acoustic emission technology, a new scheme based on adaptive improvement variational mode decomposition–wavelet packet transform is developed for extracting acoustic emission signal features of metallic panels. Specifically, three different dimensions of Q235B steel plates are utilized to collect acoustic emission signal during three-point bending experiments, to evaluate the effectiveness of the proposed approach and to investigate the influence of size effect on the acoustic emission signal characteristics. In addition, the transient process and centroid frequency distribution of each damage stage are investigated, and the internal structure variations in the bending damage process are detected by scanning electron microscopy inspection. Moreover, the generalization of the proposed damage monitoring method is evaluated for plate-like structures that have complex geometric features, such as welds. The results demonstrate the effectiveness of the proposed method for acoustic emission–based structural health monitoring of metallic plate-like structures.

Multichannel intelligent fault diagnosis of hoisting system using differential search algorithm‐variational mode decomposition and improved deep convolutional neural network

“…VMD is a completely non‐recursive decomposition algorithm, in which the modes are extracted simultaneously, and it has been widely applied in AE signal feature extraction 9 . As is well‐known, mode number K and balance parameter a play a key role in VMD, but it is difficult to empirically determine the two parameters.…”

“…VMD is a completely non-recursive decomposition algorithm, in which the modes are extracted simultaneously, and it has been widely applied in AE signal feature extraction. 9 As is well-known, mode number K and balance parameter a play a key role in VMD, but it is difficult to empirically determine the two parameters. Previous studies have shown that the mode bandwidth decreases as a increases, and the mode with large bandwidth may contain excessive interference, while the mode with small bandwidth may weaken the fault information.…”

“…To improve the generalization and prevent overfitting, the small-scale convolutional layer of the proposed IDCCN can be defined by the following Equation (9).…”

“…As a dynamic nondestructive detection method, the dynamic stress and damage degree of material can be determined effectively according to AE parameters (e.g. AE amplitude, counts and energy accumulation) 8–10 . Specifically, an important characteristic of AE technology is that AE signals generally generate by damage defects, among which crack propagation is commonly regarded as signal source 11,12 .…”

“…AE amplitude, counts and energy accumulation). [8][9][10] Specifically, an important characteristic of AE technology is that AE signals generally generate by damage defects, among which crack propagation is commonly regarded as signal source. 11,12 In general, AE-based fault diagnosis approach mainly concentrates on two stages (i.e., AE feature extraction and source localization).…”