Application Cases

Experiment Name: Analysis of Guided Wave Array Signals in Single-Layer Aluminum Plates and Stiffened Plates Using SLDV

Research Direction: Imaging analysis of debonding defects in laminated plate structures using ultrasonic waves

Experiment Objective: To collect guided wave array signals in single-layer aluminum plates and stiffened plates using SLDV, and to construct an experimental platform for piezoelectric excitation/scanning laser Doppler vibrometer reception sensing as shown in the figure below.

Test Equipment: Computer, SLDV, signal generator, power amplifier, piezoelectric patches, and vibration isolation table

Experimental Procedure:

Figure: Experimental Platform

To achieve better experimental detection results, improve the accuracy of collected signals, and reduce the impact of environmental noise, the entire experimental platform was set up on a precision vibration isolation table. Reflective tape was applied to the scanning area of the instrument to enhance focusing and reflection effects.

In the experiment, the single-layer aluminum plate specimen measured 400 mm × 400 mm × 1 mm, as shown in Figure a below. The damage was set as double through-holes with a radius of 5 mm, randomly positioned. Eleven PZT patches with a center-to-center spacing of 10 mm were attached to the upper surface of the plate to generate ultrasonic guided waves. The excitation array was distributed parallel to the y-axis at x = 25 mm, covering a range of [-55, 55] mm. To generate array data, scanning points were distributed parallel to the y-axis at x = 40 mm, covering a range of [-55, 55] mm, with an interval of 0.726 mm, resulting in a total of 139 scanning points. The corresponding guided wave responses were collected using a laser Doppler vibrometer. As shown in the figure below, excitation was performed at each PZT location sequentially. A 5-cycle windowed excitation signal with a center frequency of 100 kHz was generated by a signal generator and amplified to 150 V using a power amplifier. The sampling rate used in the experiment was 5.12 MHz, with a sampling time of 400 μs.

Figure: Stiffened Plate Specimen

In the experiment, the stiffened plate specimen measured 400 mm × 400 mm × 1 mm, as shown in Figure b above. An aluminum plate was used as the base, and stiffeners were bonded at the center of the aluminum plate. Teflon resin tape was placed between the bonding surfaces of the stiffeners and the aluminum plate to simulate damage. Eleven PZT patches with a center-to-center spacing of 10 mm were attached to the upper surface of the plate to generate ultrasonic guided waves. All other experimental settings and parameters were consistent with those of the single-layer plate experiment.

Experimental Results:

The figure below shows the wavefield diagrams at different times calculated using COMSOL simulation software, with the excitation point at location 1. Figure a shows the wavefield at 30 μs, where the Lamb wave has not yet reached the damage location. Figure b shows the wavefield at 54 μs, where the absorbing layer settings allow the Lamb wave to be rapidly absorbed and attenuated upon reaching the boundary, greatly reducing interference from boundary reflections on the damage scattering signals. Figure c shows the wavefield at 80 μs, where the Lamb wave reaches the damage location and interacts with the damage boundary, forming a scattered wavefield. From Figure c, it can be observed that the scattering signal from the damage is very weak when viewed from the steel side. Therefore, a similar wavefield diagram was generated from the rubber side for comparison, as shown in Figure d.

From the above discussion, it is evident that the scattering signal at the damage location in the steel-rubber laminated plate model is relatively weak. Therefore, although the boundary absorbing layer significantly reduces the impact of boundary reflections, it still interferes to some extent with the damage scattering signals, thereby affecting the imaging results. As shown in Figure c above, boundary echo signals generally arrive at the receiving array earlier than damage echo signals. Such echo interference can often be addressed by subtracting the baseline signal of a healthy plate, enabling higher-resolution imaging. Additionally, due to dispersion effects during Lamb wave propagation, the waveform exhibits noticeable spreading, which becomes more severe with increasing propagation distance. To effectively mitigate the impact of dispersion and achieve precise damage imaging, this study proposes a method for super-resolution imaging of damage using deep learning models. The results and analysis are presented in subsequent chapters.

For a single through-hole damage, the TFM imaging results based on numerical simulation and experiments are shown in Figures b and c above, respectively. For comparison, the numerical simulation imaging results based on the phased array imaging method are also provided, as shown in Figure a. Comparing Figures a and b, it can be observed that the TFM method provides multidimensional damage scattering information (including reflection amplitude and wave arrival time) about the damage shape and location, compared to the phased array imaging method. Comparing Figures a and c, it can be seen that the TFM-based experimental results also exhibit relatively high imaging accuracy. Therefore, the TFM imaging results from both numerical simulations and experimental data are used to configure the deep learning network database.

Figure: Specifications of the ATA-8000 Series RF Power Amplifier