Wavelet best basis compressed sensing of ultrasonic guided waves

A novel Compressed Sensing (CS) procedure is presented in this study for dispersive guided wave propagation analysis in passive structure health monitoring applications. The proposed approach combines theWavelet Packet multiresolution analysis, best basis selection and coefficients thresholding to generate a sparse but accurate time-frequency representation of the acquired dispersive signal, with the CS framework to efficiently compress Lamb waves signals. This approach is tested on experimental data obtained by passive excitation in a 1 m square aluminum plate and acquiring the dispersive signal with a conventional piezoelectric sensor. The proposed algorithm performance is analysed in term of compression ratio and percent residual difference. Results show the improvement in signal reconstruction with the use of the modified CS framework respect to the EZW coding, and the robustness of the proposed approach to additive noise in transmission.