In-situ optimized PWAS phased arrays for Lamb wave structural health monitoring

Abstract

The use of piezoelectric wafer active sensors (PWAS) phased arrays for Lamb wave damage detection in thin-wall structures is presented. The PWAS capability to tune into specific Lamb-wave modes (which is an enabling factor for our approach) is first reviewed. Then, a generic beamforming formulation that does not require the conventional parallel-ray approximation is developed for PWAS phased arrays in connection with the delay-and-sum beamforming principles. This generic formulation is applied to a 1-D linear PWAS phased array. Particularly, 1-D PWAS array beamforming reduces to the simplified parallel ray algorithm when the parallel ray approximation is invoked. The embedded ultrasonic structural radar (EUSR) algorithm is presented. A couple of simple experiments are used to show that the linear EUSR PWAS phased array system can successfully detect cracks in large aluminum thin plates. To improve the EUSR image quality, advanced signal processing is studied for possible integration into the EUSR system. The approaches include Hilbert transform for envelope detection, thresholding techniques for removing background noise, discrete wavelet transform for denoising, continuous wavelet transform for single frequency component extraction, and cross-correlation for time-of-flight detection. The optimization of linear PWAS arrays is studied next. First we consider the effect of several parameters affecting the phased-array beamforming: (1) number of elements M; (2) elementary spacing d; (3) steering angle 0; (4) location of the target r. Second, we examine the so-called nonuniform PWAS arrays which are generated by assigning different excitation weights to each of the array elements. The design of two nonuniform linear PWAS arrays, the binomial array and the Dolph‐Chebyshev array, is presented. Significant improvement of the EUSR image is observed when using these nonuniform arrays.