An Optimization Method for Acoustic Impedance Estimation of Layered Structures Using Prior Knowledge

Abstract

We present an optimization method for estimating acoustic impedance profiles of layered composite materials from ultrasonic pulse-echo data when some prior knowledge is available. The method assumes that: (1) the defect-free material consists of a small number of layers with approximately known thicknesses, impedances, and frequency-independent attenuations, and (2) the defects are thin and consist of either disbonds at an interface or delaminations within a layer. Using the prior knowledge of the impedances as an initial estimate, the impedances may be optimized for a particular set of layer thicknesses. The structure of the layers may then be adjusted, in ways consistent with the nature of expected defects, to improve the fit to the trace. These include altering the thicknesses of the layers and allowing additional layers in the region of an interface and within an existing layer. Furthermore, the impedance values within each layer may be constrained to lie within specified bounds, thus ensuring a solution consistent with the known physical structure of the material. The method may be applied to noisy and band-limited data, and may also be formulated so as to allow recursive estimation of impedance profiles in the absence of prior knowledge. Illustrations of its performance are given in examples using synthetic and real data.