Compressed sensing–based electromechanical admittance data loss recovery for concrete structural health monitoring

Abstract

Considerable amount of electromechanical admittance data needs to be collected, transmitted and stored during in-situ and long-term structural health monitoring applications, and data loss could be inevitably met when processing the monitoring electromechanical admittance signals. In this article, an innovative compressed sensing–based approach is proposed to implement data recovery for electromechanical admittance technique–based concrete structural health monitoring. The basis of this approach is to first project the original conductance signature onto an observation vector as sampled data, and then transmit the observation vector with data loss to storage station, and finally recover the missing data via a compressed sensing process. For comparison, both convex optimization theory and orthogonal matching pursuit algorithm are introduced to accomplish the compressed sensing–based electromechanical admittance data loss recovery. Prior detection test of a concrete cube subjected to varied temperatures and practical monitoring experiment of full-scale concrete shield tunnel segment undergone bolt-loosened defects are utilized to validate the feasibility of the proposed approach. In lost electromechanical admittance data recovery process, two types of data loss, namely, single-consecutive-segment loss and multiple-consecutive-segment losses, in sampled data are taken into consideration for sufficiently interpreting the effectiveness and accuracy of the convex optimization and orthogonal matching pursuit approaches. In the temperature recognition and damage identification stage, amplitude and frequency shifts in resonance peaks, cooperated with a common statistical index called root-mean-squared-deviation, are harnessed to achieve the goal after the lossy conductance signatures are recovered. The results show that the orthogonal matching pursuit–based data recovery approach is superior to the convex optimization approach because of its smaller calculation of consumption as well as lower recovered errors.