High-resolution inversion of ultrasonic traces

Abstract

An algorithm for estimating the acoustic reflection coefficient profile from ultrasonic traces obtained during inspection of layered materials is described. Given the measured trace and the incident wavelet, the inversion proceeds by means of a layer stripping approach combined with high-resolution deconvolution. The inversion algorithm is stable to noise and is suitable for use with bandlimited data. It is particularly suitable for use with materials that exhibit a few large discontinuities in impedance and in which multiple reflections in the data are evident. The performance of the algorithm is illustrated in tests with synthetic and real data. An implementation of the algorithm on a TMS 320C30 signal processing board allowed the inversion of an entire set of 256 traces, each of 256 elements, in 15 s.

An Inversion Approach to Ultrasonic Imaging through Reflective Interfaces in Multi-Layered Structures

“…(a) B-scan generated from data following reference subtraction (b) C-scan generated from data following reference subtraction been developed [3,4). This algorithm has been described elsewhere [3) and will not be presented in detail here. In brief, however, a residual trace is iteratively computed by first identifying potential reflection coefficients, updating a forward modelled trace and calculating the residual.…”

“…Recently, an efficient inversion algorithm based on the combined application of a layer stripping approach and L2 norm deconvolution has Figure 7. (a) B-scan generated from data following reference subtraction (b) C-scan generated from data following reference subtraction been developed [3,4). This algorithm has been described elsewhere [3) and will not be presented in detail here.…”

Application of High Resolution Inversion of Ultrasonic Data to the Imaging of Multi-Layered Composite Structures

“…[7-l0] The "exact" methods provide the analytically exact solution, but require that the input data be totally broadband and noise free. [11] Neither of these two criteria are ever met in practical NDE applications, which encounter several noise sources and a bandlimited wavelet. Therefore, exact inversion techniques are seldom directly applicable to these problems.…”

“…[9-10) Each of the estimation techniques has some practical value, although all suffer from some practical limitations imposed by the nature of NDE ultrasonic data. For example, some are unable to accornrnodate the arbitrary wavelets generated during pulse-echo ultrasonic imaging, [11) while others are very sensitive to the amount of noise prevalent in most NDE ultrasonic data.…”

“…[11,12) Recently, a high resolution inversion algorithm which employs an optimization approach has been proposed by Zala . [12 , 13) This algorithm also makes use of several features that have been used in other algorithms which are less successful for NDE applications, such as:…”