Deformation Data Recovery Based on Compressed Sensing in Bridge Structural Health Monitoring

Abstract

Full life-cycle and real-time structural health monitoring (SHM) rely on numerous, heterogeneous sets of data collected from sensors and extracting feathers that support the estimation of the health condition of a bridge. It is the principal challenges facing the SHM application on transmission and storing such huge amounts of data. Compressive sensing (CS) is a novel data acquisition method whereby the compression is done in a sensor simultaneously with the sampling. In this work, we established the possibility of compressed sensing to address this challenges on bridge SHM. Based on the sparsity of the deformation data of bridge, we proposed a random sampling scheme based on CS to minimize the number of field data. The CS recovery performance is mostly determined by the decomposition basis which is associated with the sparsity of the sampling signal. Different bases have been tested to recover the deformation data. Experimental results demonstrate the proposed method allows a reduction of the measurement data with an acceptable recovery accuracy. And reconstruction performance based on DWT to sparse transform is better than that based on DCT to sparse transform. When the compression ratio is above 0.6, the reconstruction error grows moderately; whereas the reconstruction error grows rapidly as the compression ratio is below 0.6. With the compression ratio decreased from 0.6 to 0.2, reconstruction error is reduced about 2.5 times by using DWT and reduced about 2 times by using DCT.