Compressed sensing in concrete pipes with elastic waves

Abstract

This paper presents compressed sensing (CS) for non-destructive characterization of concrete pipes utilizing elastic guided waves. In the conventional multichannel analysis of surface wave method (MASW), unknown geometric and elastic properties of a pipe are estimated by mathematically inverting dispersion curves measured over the pipe. While this is a well-established technique, several practical challenges exists during implementation. For example, wave record is captured using a densely packed receiver array to satisfy the Nyquist sampling theorem, which leads to large element count. This paper develops CS based signal acquisition strategy using sparse sub-Nyquist array that is randomly arranged over the pipe. Results show that high quality multi-modal dispersion images of different pipe modes could be recovered from such sparse datasets. Finally, the modal curves are mathematically inverted using wave theory based forward models, and the unknown parameters of the pipe are calculated. The results are validated using data obtained using an independent NDT method. The proposed approach is verified using longitudinal $L(n,m)$ modes and circumferential Lamb type modes ($CLT$) modes. Investigations are conducted using synthetic and measured datasets. Overall, the proposed approach enhances the conventional MASW technique and improves productivity.