Ultrasonic NDT Signal Acquisition and Processing (USAP) System based on ZYNQ APSoC (2015-2019)

Introduction

  • Ultrasonic nondestructive testing (NDT) is an effective and highly practical method to characterize thickness and check internal structure of the test subject with a high precision.

  • In this study, we designed and implemented a reconfigurable, high performance and low cost ultrasonic NDE platform based on Xilinx ZYNQ all-programmable System-on-Chip (APSoC).

  • This system can generate high frequency ultrasonic pulse and capture the backscattered echoes at the frequency of 250 MSPS.

  • Applications of ultrasonic flaw detection based on Multilayer Perceptron Neural Network (MLPNN) is introduced for this study. Time domain ultrasonic backscattered signal and its Split Spectrum Processing (SSP) 2D representation are used for training the MLPNN.

System Implementation

../../_images/USAP_bd.PNG

Figure. 1. USAP System Block Diagram.

../../_images/USAP_sysimplementation.PNG

Figure. 2. USAP System Implementation.

../../_images/USAP_signalprocess.PNG

Figure. 3. Signal Processing on USAP system.

Example Result

Figure 4. is the slice plane plot of the envelopes of the ultrasonic data. Before plotting, Hilbert filters are applied to the ultrasonic A-scans to extract envelops of the signals. This will remove the high frequency oscillations in the ultrasonic A-scans and make the amplitude variations more obvious in the plots.

../../_images/USAP_testresult.PNG

Figure. 4. Slice Plane Plot of the Test Results.

For demonstration, I applied 2D NDT scanning to six different steel blocks (labeled by Sample 0 to Sample 5) to collect backscattered signals. Each 2D NDT scanning result collected contains \(200 \times 200\) A-scans along the x-axis and y-axis, covering a \(2 \times 2 cm^{2}\) area of the test specimens. Rendering the 3D ultrasonic data in different methods can reveal different aspects of information in the test specimen. We will provide four different volumetric data visualization methods for demonstration, including slice plane plot, volume rendering, max intensity project, and isosurface plot.

Figure 5 shows the slice plane plots of the envelopes of collected datasets. Each slice plane plot contains three parts, including the XY plane plot, XZ plane plot, and XY plane plot. These plotted planes are selected from the middle of the volumetric data, as it is shown in Figure 5. The slice plane plot provides intuitive views of B-scans and C-scans of the volumetric data.

../../_images/2_15.PNG

Figure. 5. Slice Plane Plots of the Ultrasonic Volumetric Data Envelopes.

Figure 6 is the maximum intensity projection (MIP) plots of the collected ultrasonic datasets. MIP is a method for volumetric data visualization that projects the voxels with the highest intensity to the 2D visualization planes. As shown in Figure 6, MIP successfully enhances the high-intensity echoes in the ultrasonic dataset. It helps the researchers to locate the potential abnormal echoes in the dataset.

../../_images/2_16.PNG

Figure. 6. MAX Intensity Projection of the Ultrasonic Volumetric Data.

Figure 7 shows the volume rendering plots of the envelopes of collected datasets. An alphamap is necessary for the volume rendering plots to define the opacity of the voxels according to the intensity. The volume rendering plots in Figure 7 utilize the linear alphamap.

../../_images/2_17.PNG

Figure. 7. Volume Rendering.

Figure 8 shows the isosurface plots of the envelopes of collected ultrasonic datasets. Different from other volumetric data visualization methods, the isosurface of the 3D dataset can be acquired by plotting the boundaries of the voxels with the same intensity. This constant value of intensity can be specified by the researchers according to the dataset to be displayed. Isosurface plots in Figure 8 can successfully reveal the textures of the collected data in a 3D space.

../../_images/2_18.PNG

Figure. 8. Isosurface Plots of the Ultrasonic Volumetric Data Envelopes.

Note

Note: The page is a demonstration of the project basic layouts, details can be found in my papers.


PUBLICATIONS

  1. Pramod Govindan, Boyang Wang, Pingping Wu, Ivan Palkov, Vidya Vasudevan, and Jafar Saniie. Reconfigurable and programmable system-on-chip hardware platform for real-time ultrasonic testing applications. In 2015 IEEE International Ultrasonics Symposium (IUS), 1–4. IEEE, 2015. doi:10.1109/ULTSYM.2015.0341.

  2. Pramod Govindan, Boyang Wang, Prashaanth Ravi, and Jafar Saniie. Hardware and software architectures for computationally efficient three-dimensional ultrasonic data compression. IET Circuits, Devices & Systems, 10(1):54–61, 2016. doi:10.1049/iet-cds.2015.0083.

  3. Boyang Wang, Pramod Govindan, and Jafar Saniie. Performance analysis of system-on-chip architectures for ultrasonic data compression. In 2016 IEEE International Ultrasonics Symposium (IUS), 1–4. IEEE, 2016. doi:10.1109/ULTSYM.2016.7728507.

  4. Vidya Vasudevan, Boyang Wang, Pramod Govindan, and Jafar Saniie. Design and evaluation of reconfigurable ultrasonic testing system. In 2015 IEEE International Conference on Electro/Information Technology (EIT), 310–313. IEEE, 2015. doi:10.1109/EIT.2015.7293359.

  5. Boyang Wang, Pramod Govindan, Thomas Gonnot, and Jafar Saniie. Acceleration of ultrasonic data compression using opencl on gpu. In 2015 IEEE International Conference on Electro/Information Technology (EIT), 305–309. IEEE, 2015. doi:10.1109/EIT.2015.7293358.

  6. Boyang Wang. Reconfigurable Ultrasonic Signal Processing System Solution Based on Zynq Platform. PhD thesis, Illinois Institute of Technology, 2015.

  7. Boyang Wang and Jafar Saniie. A high performance ultrasonic system for flaw detection. In 2019 IEEE International Ultrasonics Symposium (IUS), 840–843. IEEE, 2019. doi:10.1109/ULTSYM.2019.8926280.