Singular spectrum analysis and fuzzy entropy-based damage detection on a thin aluminium plate by using PZTs

Abstract

In this research, a new method based on singular spectrum analysis (SSA) and fuzzy entropy is developed for damage detection on thin wall-like structures, and the normalized fuzzy entropy is employed as an indicator to identify the severity of the damage. The Lead Zirconate Titanate (PZT) transducers are used in this research to generate and detect the Lamb waves. During the detection, the collected signals from the PZT sensors are firstly decomposed and reconstructed by SSA to extract the feature of the damage, and then the reconstructed signals with the feature of the damage are processed to obtain the normalized fuzzy entropy. An experimental setup of an aluminium plate with added magnets is fabricated to validate the proposed method. The experimental results show that when magnets are attached on the aluminium plate, the normalized fuzzy entropy is smaller than that when there are no magnets. That is because when magnets are placed on the plate, the movement and some vibration modes of Lamb waves are disturbed by the added magnets and this disturbing effect can be enhanced by increasing the number and locations of the added magnets, and eventually the complexity and nonlinearity of the waves are weakened. The experimental results of a single damage with different number of magnets indicate that the normalized fuzzy entropy decreases linearly as the number of the added magnets increases, which demonstrates that the proposed method can be used to detect the severity of the damage. Moreover, the experimental results of multi-damage on different locations indicate that the normalized fuzzy entropy is relevant with both the total number and locations of the added magnets. The normalized fuzzy entropy decreases linearly as the total number of the magnets increases, and the entropy of a single damage is smaller than that of the multi-damage with the same total number of magnets, which demonstrates that the proposed method also can be used for multi-damage detection on a thin plate. This study provides us a new approach to identifying a single or multiple damages on thin wall-like structures.

Underwater bolted flange looseness detection using percussion-induced sound and Feature-reduced Multi-ROCKET model

“…Specifically, eight datasets are collected by two researchers in 1 week and operators first loosen the bolts before tightening them under each preload level in each dataset, which ensures that classes are independent of each other and datasets are independent of each other. Datasets (1, 2, 3, 4) are taken from the flange A and datasets (5,6,7,8) are taken from the flange B. Figure 6 exhibits single-hit audio signals under different datasets and preload levels.…”

“…Derived from the typical 1D convolutional kernel in the convolutional neural network, ROCKET 42 introduces a very large number of 1D convolutional kernels, which have random and different length, bias, dilation, weights, and paddings, to capture feature maps for the input time series. Particularly, the length of each kernel is selected randomly from three values (7,9,11) given the same probability. In addition, the values of weights are sampled from a normal distribution and the values of biases are sampled from a uniform distribution.…”

“…In the past decades, benefitting from the rapid development in structural health monitoring (SHM), [5][6][7] researchers have been making contributions to the detection of bolted connection looseness in the air. [8][9][10] Notably, detection approaches based on the piezoelectric material 11 have been developed.…”

Nonlinear Vibration Analysis of Beam and Plate with Closed Crack: A Review

“…Non-zero energy flux and phase velocity are required due to the dispersion and the multimodal character of Lamb waves. As a result, different researchers [71] have chosen different mode pairs for different damages, although the amplitude of the required second harmonic Lamb mode was always altered by other Lamb waves with double frequency [72]. For Lamb waves in partially closed cracks, W. Zhu et al [73] use a static component.…”

Damage Diagnosis in Circular Structures using Cartesian Wavelet Analysis: A Comparison between Two Structural Signals

“…The performance of the method was assessed by performing several experiments on an aluminum alloy plate using four piezoceramics to act as exciters and receivers, and the Lamb waves to scan the plate structure. Song et al [18] used the fuzzy entropy and singular spectrum analysis to establish an innovative algorithm for damage detection in thin plates. To generate and detect the Lamb waves, the lead zirconate titanate transducers were employed.…”