Alan J. Hunter

AIP Conference Proceedings

The quality of an ultrasonic array image, especially for anisotropic material, depends on accurate information about acoustic properties. Inaccuracy of acoustic properties causes image degradation, e.g., blurring, errors in locating of reflectors and introduction of artifacts. In this paper, for an anisotropic austenitic steel weld, an autofocus imaging technique is presented. The array data from a series of beacons is captured and then used to statistically extract anisotropic weld properties by using a Monte-Carlo inversion approach. The beacon and imaging systems are realized using two separated arrays; one acts as a series of beacons and the other images these beacons. Key to the Monte-Carlo inversion scheme is a fast forward model of wave propagation in the anisotropic weld and this is based on the Dijkstra algorithm. Using this autofocus approach a measured weld map was extracted from an austenitic weld and used to reduce location errors, initially greater than 6mm, to less than 1mm.

The wavenumber algorithm for full-matrix imaging using an ultrasonic array

Ultrasonic imaging using full-matrix capture, e.g., via the total focusing method (TFM), has been shown to increase angular inspection coverage and improve sensitivity to small defects in nondestructive evaluation. In this paper, we develop a Fourier-domain approach to full-matrix imaging based on the wavenumber algorithm used in synthetic aperture radar and sonar. The extension to the wavenumber algorithm for full-matrix data is described and the performance of the new algorithm compared with the TFM, which we use as a representative benchmark for the time-domain algorithms. The wavenumber algorithm provides a mathematically rigorous solution to the inverse problem for the assumed forward wave propagation model, whereas the TFM employs heuristic delay-and-sum beamforming. Consequently, the wavenumber algorithm has an improved point-spread function and provides better imagery. However, the major advantage of the wavenumber algorithm is its superior computational performance. For large arrays and images, the wavenumber algorithm is several orders of magnitude faster than the TFM. On the other hand, the key advantage of the TFM is its flexibility. The wavenumber algorithm requires a regularly sampled linear array, while the TFM can handle arbitrary imaging geometries. The TFM and the wavenumber algorithm are compared using simulated and experimental data.

Monte carlo inversion of ultrasonic array data to map anisotropic weld properties

The quality of an ultrasonic array image depends on accurate information about its acoustic properties. Inaccurate acoustic properties can cause image degradation such as blurring, mislocation of reflectors, and the introduction of artifacts. In this paper, for the specific case of an inhomogeneous and anisotropic austenitic steel weld, Monte Carlo Markov Chain (MCMC) inversion is used to estimate unknown acoustic properties from array data. The approach uses active beacons that transmit ultrasound through the anisotropic weld; the ultrasound is then captured by a receiving array. A forward model of the ultrasonic array data is then optimized with respect to the experimental data using an MCMC inversion. The result of this process is the extraction of a material property map that describes the anisotropy distribution within the weld region. These extracted material properties are then used within an imaging algorithm - the total focusing method in this paper - to produce autofocused images. This MCMC inversion approach is first applied to simulated data to test the convergence, robustness, and accuracy of the method and its implementation. The extracted weld map is used to show improved imaging of defects within the weld, relative to an image formed assuming a constant velocity. Finally, the MCMC inversion approach is used on experimental data from a 110-mm-thick steel plate containing an austenitic weld. Here the extracted weld map is used to show that defect location errors of greater than 5 mm are reduced to around 2 mm when the extracted weld map is used.

3-D reconstruction of sub-wavelength scatterers from the measurement of scattered fields in elastic waveguides

In nondestructive testing, being able to remotely locate and size defects with good accuracy is an important requirement in many industrial sectors, such as the petrochemical, nuclear, and aerospace industries. The potential of ultrasonic guided waves is well known for this type of problem, but interpreting the measured data and extracting useful information about the defects remains challenging. This paper introduces a Bayesian approach to measuring the geometry of a defect while providing at the same time an estimate of the uncertainty in the solution. To this end, a Markov-chain Monte Carlo algorithm is used to fit simulated scattered fields to the measured ones. Simulations are made with efficient models where the geometries of the defects are provided as input parameters, so that statistical information on the defect properties such as depth, shape, and dimensions can be obtained. The method is first investigated on simulations to evaluate its sensitivity to noise and to the amount of measured data, and it is then demonstrated on experimental data. The defect geometries vary from simple elliptical flat-bottomed holes to complex corrosion profiles.

Passive acoustic detection of closed-circuit underwater breathing apparatus in an operational port environment.

Divers constitute a potential threat to waterside infrastructures. Active diver detection sonars are available commercially but present some shortcomings, particularly in highly reverberant environments. This has led to research on passive sonar for diver detection. Passive detection of open-circuit UBA (underwater breathing apparatus) has been demonstrated. This letter reports on the detection of a diver wearing closed-circuit UBA (rebreather) in an operational harbor. Beamforming is applied to a passive array of 10 hydrophones in a pseudo-random linear arrangement. Experimental results are presented demonstrating detection of the rebreather at ranges up to 120 m and are validated by GPS ground truth.

Least-squares estimation of imaging parameters for an ultrasonic array using known geometric image features

Ultrasonic array images are adversely affected by errors in the assumed or measured imaging parameters. For non-destructive testing and evaluation, this can result in reduced defect detection and characterization performance. In this paper, an autofocus algorithm is presented for estimating and correcting imaging parameter errors using the collected echo data and a priori knowledge of the image geometry. Focusing is achieved by isolating a known geometric feature in the collected data and then performing a weighted least-squares minimization of the errors between the data and a feature model, with respect to the unknown parameters. The autofocus algorithm is described for the estimation of element positions in a flexible array coupled to a specimen with an unknown surface profile. Experimental results are shown using a prototype flexible array and it is demonstrated that (for an isolated feature and a well-prescribed feature model) the algorithm is capable of generating autofocused images that are comparable in quality to benchmark images generated using accurately known imaging parameters.

Sonar target enhancement by shrinkage of incoherent wavelet coefficients

Background reverberation can obscure useful features of the target echo response in broadband low-frequency sonar images, adversely affecting detection and classification performance. This paper describes a resolution and phase-preserving means of separating the target response from the background reverberation noise using a coherence-based wavelet shrinkage method proposed recently for de-noising magnetic resonance images. The algorithm weights the image wavelet coefficients in proportion to their coherence between different looks under the assumption that the target response is more coherent than the background. The algorithm is demonstrated successfully on experimental synthetic aperture sonar data from a broadband low-frequency sonar developed for buried object detection.

Registration of images from a hull mounted, low frequency synthetic aperture sonar

Coherent change detection between multiple synthetic aperture sonar (SAS) images is reliant on the images being co-registered with sub-pixel accuracy. In this paper we suggest a technique using available navigation data to reconstruct the images onto a common grid. Data obtained using the MUD SAS system is used to demonstrate this method. We show that even with high quality navigation data there is still a misregistration which causes a loss of coherence. Additional data-driven alignment is required in order to use the images for coherent change detection.

MONTE-CARLO INVERSION OF TRAVEL-TIME DATA FOR THE ESTIMATION OF WELD MODEL PARAMETERS

The quality of ultrasonic array imagery is adversely affected by uncompensated variations in the medium properties. A method for estimating the parameters of a general model of an inhomogeneous anisotropic medium is described. The model is comprised of a number of homogeneous sub‐regions with unknown anisotropy. Bayesian estimation of the unknown model parameters is performed via a Monte‐Carlo Markov chain using the Metropolis‐Hastings algorithm. Results are demonstrated using simulated weld data.

A study into the effects of an austenitic weld on ultrasonic array imaging performance

An industrial application of ultrasonic array imaging is the inspection of austenitic welds with high inhomogeneity and anisotropy. These result in attenuation and perturbation of the signals that adversely affects imaging performance. Here, the effects of perturbations introduced by an austenitic weld on array imaging performance are investigated experimentally. It is shown that three major factors contribute to the degradation of image quality: timing errors, phase errors, and multi‐path propagation and scattering.