Session: 04 - 01 Emerging Techniques & Technology

Paper Number: 118334

118334 - Noninvasive Acoustic Time-of-Flight (Tof) Measurement Using Convoloutional Neural Network 

Acoustic Time-of-Flight (ToF) techniques for assessing fluid-filled containers/pipes are used in critical engineering industries, including monitoring nuclear reactors, chemical processing, and production of oil and gas. However, quantifying ToF of bulk waves in fluid-filled cylindrical container pose challenges to accurately predicting ToF because guided waves in the cylinder walls interfere with the bulk waves in the fluid. In this work, a data-driven approach based on Convolutional Neural Networks (CNN) proposed by Greenhall et al. is implemented to quantify the ToF of bulk waves on various size cylindrical containers. The CNN technique can assist in identifying critical features in the measured waveforms and recognize bulk wave arrivals, which is challenging using conventional techniques. We investigate the generalizability of the CNN technique, in terms of how accurately the ToF can be measured in situations when container dimensions and acoustic excitation signals differ between the test and training data. This represents the real-world situation where the ToF measurement technique could be implemented on any fluid-filled container even if the CNN has not been trained on data with the same container dimensions and acoustic excitation. The network performance is characterized as a function of relative error. The results show the potential of CNNs for identifying ToF in critical engineering components to improve reliability. The ToF data can be further use in acoustic imaging techniques for image reconstruction.

Presenting Author: Abhishek Saini Los Alamos National Laboratory

Presenting Author Biography: Abhishek Saini is a postdoc research associate at Los Alamos National Laboratory. He completed his PhD from Nanyang Technological University Singapore in 2021 and worked as Postdoctoral Fellow with RollsRoyce@NTU corporate laboratory. His research interests include acoustics, ultrasound, nondestructive evaluation, imaging and inversion, metamaterials and applied machine learning.