Signal denoising and ultrasonic flaw detection via overcomplete and sparse representations

Flaw Detection in Highly Scattering Materials Using a Simple Ultrasonic Sensor Employing Adaptive Template Matching

Ultrasonic sensors have been extensively used in the nondestructive testing of materials for flaw detection. For polycrystalline materials, however, due to the scattering nature of the material, which results in strong grain noise and attenuation of the ultrasonic signal, accurate detection of flaws is particularly difficult. In this paper, a novel flaw-detection method using a simple ultrasonic sensor is proposed by exploiting time-frequency features of an ultrasonic signal. Since grain scattering mostly happens in the Rayleigh scattering region, it is possible to separate grain-scattered noise from flaw echoes in the frequency domain employing their spectral difference. We start with the spectral modeling of grain noise and flaw echo, and how the two spectra evolve with time is established. Then, a time-adaptive spectrum model for flaw echo is proposed, which serves as a template for the flaw-detection procedure. Next, a specially designed similarity measure is proposed, based on which the similarity between the template spectrum and the spectrum of the signal at each time point is evaluated sequentially, producing a series of matching coefficients termed moving window spectrum similarity (MWSS). The time-delay information of flaws is directly indicated by the peaks of MWSSs. Finally, the performance of the proposed method is validated by both simulated and experimental signals, showing satisfactory accuracy and efficiency.

Sparse and Dispersion-Based Matching Pursuit for Minimizing the Dispersion Effect Occurring when Using Guided Wave for Pipe Inspection

Ultrasonic guided wave is an effective tool for structural health monitoring of structures for detecting defects. In practice, guided wave signals are dispersive and contain multiple modes and noise. In the presence of overlapped wave-packets/modes and noise together with dispersion, extracting meaningful information from these signals is a challenging task. Handling such challenge requires an advanced signal processing tool. The aim of this study is to develop an effective and robust signal processing tool to deal with the complexity of guided wave signals for non-destructive testing (NDT) purpose. To achieve this goal, Sparse Representation with Dispersion Based Matching Pursuit (SDMP) is proposed. Addressing the three abovementioned facts that complicate signal interpretation, SDMP separates overlapped modes and demonstrates good performance against noise with maximum sparsity. With the dispersion taken into account, an overc-omplete and redundant dictionary of basic atoms based on a narrowband excitation signal is designed. As Finite Element Method (FEM) was used to predict the form of wave packets propagating along structures, these atoms have the maximum resemblance with real guided wave signals. SDMP operates in two stages. In the first stage, similar to Matching Pursuit (MP), the approximation improves by adding, a single atom to the solution set at each iteration. However, atom selection criterion of SDMP utilizes the time localization of guided wave reflections that makes a portion of overlapped wave-packets to be composed mainly of a single echo. In the second stage of the algorithm, the selected atoms that have frequency inconsistency with the excitation signal are discarded. This increases the sparsity of the final representation. Meanwhile, leading to accurate approximation, as discarded atoms are not representing guided wave reflections, it simplifies extracting physical meanings for defect detection purpose. To verify the effectiveness of SDMP for damage detection results from numerical simulations and experiments on steel pipes are presented.

A sparse digital signal model for ultrasonic nondestructive evaluation of layered materials

Signal modeling has been proven to be an useful tool to characterize damaged materials under ultrasonic nondestructive evaluation (NDE). In this paper, we introduce a novel digital signal model for ultrasonic NDE of multilayered materials. This model borrows concepts from lattice filter theory, and bridges them to the physics involved in the wave-material interactions. In particular, the proposed theoretical framework shows that any multilayered material can be characterized by a transfer function with sparse coefficients. The filter coefficients are linked to the physical properties of the material and are analytically obtained from them, whereas a sparse distribution naturally arises and does not rely on heuristic approaches. The developed model is first validated with experimental measurements obtained from multilayered media consisting of homogeneous solids. Then, the sparse structure of the obtained digital filter is exploited through a model-based inverse problem for damage identification in a carbon fiber-reinforced polymer (CFRP) plate.

Cluster-enhanced sparse approximation of overlapping ultrasonic echoes

Ultrasonic pulse-echo methods have been used extensively in non-destructive testing of layered structures. In acoustic measurements on thin layers, the resulting echoes from two successive interfaces overlap in time, making it difficult to assess the individual echo parameters. Over the last decade sparse approximation methods have been extensively used to address this issue. These methods employ a large dictionary of elementary functions (atoms) and attempt to select the smallest subset of atoms (sparsest approximation) that represent the ultrasonic signal accurately. In this paper we propose the cluster-enhanced sparse approximation (CESA) method for estimating overlapping ultrasonic echoes. CESA is specifically adapted to deal with a large number of signals acquired during an ultrasonic scan. It incorporates two principal algorithms. The first is a clustering algorithm, which divides a set of signals comprising an ultrasonic scan into groups of signals that can be approximated by the same set of atoms. The second is a two-stage iterative algorithm, which alternates between update of the atoms associated with each cluster, and re-clustering of the signals according to the updated atoms. Because CESA operates on clusters of signals, it achieves improved results in terms of approximation error and computation time compared with conventional sparse methods, which operate on each signal separately. The superior ability of CESA to approximate highly overlapping ultrasonic echoes is demonstrated through simulation and experiments on adhesively bonded structures.

A linear model approach for ultrasonic inverse problems with attenuation and dispersion

Ultrasonic inverse problems such as spike train deconvolution, synthetic aperture focusing, or tomography attempt to reconstruct spatial properties of an object (discontinuities, delaminations, flaws, etc.) from noisy and incomplete measurements. They require an accurate description of the data acquisition process. Dealing with frequency-dependent attenuation and dispersion is therefore crucial because both phenomena modify the wave shape as the travel distance increases. In an inversion context, this paper proposes to exploit a linear model of ultrasonic data taking into account attenuation and dispersion. The propagation distance is discretized to build a finite set of radiation impulse responses. Attenuation is modeled with a frequency power law and then dispersion is computed to yield physically consistent responses. Using experimental data acquired from attenuative materials, this model outperforms the standard attenuation-free model and other models of the literature. Because of model linearity, robust estimation methods can be implemented. When matched filtering is employed for single echo detection, the model that we propose yields precise estimation of the attenuation coefficient and of the sound velocity. A thickness estimation problem is also addressed through spike deconvolution, for which the proposed model also achieves accurate results.

Sparse signal representation and its applications in ultrasonic NDE

Many sparse signal representation (SSR) algorithms have been developed in the past decade. The advantages of SSR such as compact representations and super resolution lead to the state of the art performance of SSR for processing ultrasonic non-destructive evaluation (NDE) signals. Choosing a suitable SSR algorithm and designing an appropriate overcomplete dictionary is a key for success. After a brief review of sparse signal representation methods and the design of overcomplete dictionaries, this paper addresses the recent accomplishments of SSR for processing ultrasonic NDE signals. The advantages and limitations of SSR algorithms and various overcomplete dictionaries widely-used in ultrasonic NDE applications are explored in depth. Their performance improvement compared to conventional signal processing methods in many applications such as ultrasonic flaw detection and noise suppression, echo separation and echo estimation, and ultrasonic imaging is investigated. The challenging issues met in practical ultrasonic NDE applications for example the design of a good dictionary are discussed. Representative experimental results are presented for demonstration.