Digital Signal Processing Methods for Source Function Extraction of Piezoelectric Elements

Abstract

Guided wave techniques have great potential for the structural health monitoring of plate-like components. Previous research has demonstrated the effectiveness of combining laser-ultrasonic techniques with time-frequency representations to experimentally develop the dispersion relationship of a plate; the high fidelity, broad bandwidth and point-like nature of laser ultrasonics are critical for the success of these results. Unfortunately, laser ultrasonic techniques are time and cost intensive, and are impractical for many in-service applications. Therefore this research develops a complementary digital signal processing methodology that uses mounted piezoelectric elements instead of optical devices. This study first characterizes the spatial and temporal effects of oil coupled and glued piezoelectric sources, and then develops a procedure to interpret and model the distortion caused by their limited bandwidth and finite size. Furthermore, it outlines any inherent difficulties for time and frequency domain considerations. The deconvolution theory for source function extraction in the time - and frequency domain under the presence of noise is provided and applied to measured data. These considerations give the background for further studies to develop a dispersion relationship of a plate with the fidelity and bandwidth similar to results possible with laser ultrasonics, but made using mounted piezoelectric sources.