Image-based techniques are a promising non-destructive approach for road pavement condition evaluation. The main objective of this study is to extract, quantify and evaluate important surface defects, such as cracks, using an automated computer vision-based system to provide a better understanding of the pavement deterioration process. To achieve this objective, an automated crack-recognition software was developed, employing a series of image processing algorithms of crack extraction, crack grouping, and crack detection. Bottom-hat morphological technique was used to remove the random background of pavement images and extract cracks, selectively based on their shapes, sizes, and intensities using a relatively small number of user-defined parameters. A technical challenge with crack extraction algorithms, including the Bottom-hat transform, is that extracted crack pixels are usually fragmented along crack paths. For de-fragmenting those crack pixels, a novel crack-grouping algorithm is proposed as an image segmentation method, so called MorphLink-C. Statistical validation of this method using flexible pavement images indicated that MorphLink-C not only improves crack-detection accuracy but also reduces crack detection time. Crack characterization was performed by analysing imagerial features of the extracted crack image components. A comprehensive statistical analysis was conducted using filter feature subset selection (FSS) methods, including Fischer score, Gini index, information gain, ReliefF, mRmR, and FCBF to understand the statistical characteristics of cracks in different deterioration stages. Statistical significance of crack features was ranked based on their relevancy and redundancy. The statistical method used in this study can be employed to avoid subjective crack rating based on human visual inspection. Moreover, the statistical information can be used as fundamental data to justify rehabilitation policies in pavement maintenance. Finally, the application of four classification algorithms, including Artificial Neural Network (ANN), Decision Tree (DT), k-Nearest Neighbours (kNN) and Adaptive Neuro-Fuzzy Inference System (ANFIS) is investigated for the crack detection framework. The classifiers were evaluated in the following five criteria: 1) prediction performance, 2) computation time, 3) stability of results for highly imbalanced datasets in which, the number of crack objects are significantly smaller than the number of non-crack objects, 4) stability of the classifiers performance for pavements in different deterioration stages, and 5) interpretability of results and clarity of the procedure. Comparison results indicate the advantages of white-box classification methods for computer vision based pavement evaluation. Although black-box methods, such as ANN provide superior classification performance, white-box methods, such as ANFIS, provide useful information about the logic of classification and the effect of feature values on detection results. Such information can provide further insight for the image-based pavement crack detection application.

Additive manufacturing (AM) is a novel fabrication technique capable of producing highly complex parts. Nevertheless, a major challenge is improving the quality of the fabricated parts. While there are several ways of approaching this problem, developing data-driven methods that use AM process signatures to identify these part anomalies can be rapidly applied to improve the overall part quality during the build. The objective of this dissertation is to model multiple processes within the AM to quantify the quality of the parts and reduced the uncertainty due to variation in input process parameters. The objective of first study is to build a new layer-wise process signature model to characterize the thermal-defect relationship. Based on melt pool images, we propose novel layer-wise key process signatures, which are calculated using multilinear principal component analysis (MPCA) and are directly correlated with layer-wise quality of the part. Second study broadens the spectrum of the dissertation to include mechanical properties, where a novel two-phase modeling methodology is proposed for fatigue life prediction based on in-situ monitoring of thermal history. In final study, our objective is to pave the way toward a better understanding of the uncertainty in the process-defect-structures relationship using an inverse robust design exploration method. The method involves two steps. In the first step, mathematical models are developed to characterize and model the forward flow of information in the intended additive manufacturing process. In the second step, inverse robust design exploration is carried out to investigate satisfying design solutions that meet multiple AM goals.

No description available.

There currently exists no reliable, non-destructive method for measuring stress in railroads and other similar structures without the need for a calibration measurement. Major limitations which have hindered previous techniques include sensitivity to boundary conditions, insensitivity to stress, and intolerance for material and geometry uncertainty. In this work, a technique is developed which seeks to solve these challenges by extracting the spectrum relation, or dispersion curve, of a waveguide from dispersive wave propagation meaasurements. The technique is based on spectral analysis of waves in structures modeled as beams, and as such is based on relatively low frequency vibrations, as opposed to other techniques which use nonlinear elastic modeling of structures at ultrasonic frequencies. The major contribution of this work is the development of a frequency-domain based signal processing technique which is capable of compensating for the dispersive, long wavelength reflections which have limited the ability of previous techniques to go low enough in frequency to achieve high stress sensitivity. By compensating for reflections, the present work is able to automate the process of analyzing wave propagation signals such that the entire dispersion curve can be extracted, enabling the identification of various parameters including stress, stiffness, density, and other material and geometry properties. This in turn enables measuring stress, performing model-updating for material and geometry uncertainty, and being indifferent to boundary conditions. The theory and algorithmic implementation is presented, along with simulations and experimental validation on a rectangular beam. / Master of Science / The ability to detect damage or the potential for damage in structures is highly desirable, especially in industries such as civil infrastructure in which failure can be incredibly costly and dangerous. In particular, non-destructive techniques which can predict failure without interfering with the operations of a structure are particularly sought after. In this work, a technique for non-intrusively and non-destructively measuring stress is developed, with the primary application being for measuring stress in railroads. The technique seeks to advance the state-of-the-art in wave-propagation-based techniques by adding the capability to automatically identify reflected waves. With this new capability, the method is able to quickly and efficiently analyze a large set of vibration measurements to extract information about the structure's material, geometry, and loading characteristics which enables solving for stress even when the structures material, geometry, and boundary conditions are not precisely known. The technique is demonstrated on both simulated and experimental data, in which a rectangular beam is tensioned and the stress is then identified.

Portland cement concrete (PCC) structures deteriorate with age and need to be maintained or replaced. Early detection of deterioration in PCC (e.g., alkali-silica reaction, freeze/thaw damage, or chloride presence) can lead to significant reductions in maintenance costs. However, it is often too late to perform low-cost preventative maintenance by the time deterioration becomes evident. Non-destructive evaluation (NDE) methods are potentially among the most useful techniques developed for assessing constructed facilities. They are noninvasive and can be performed rapidly. Portland cement concrete can be nondestructively evaluated by electrically characterizing its complex dielectric constant. The real part of the dielectric constant depicts the velocity of electromagnetic waves in PCC. The imaginary part describes the conductivity of PCC and the attenuation of electromagnetic waves, and hence the losses within the PCC media. Dielectric properties of PCC have been investigated in a laboratory setting using a parallel plate capacitor operating in the frequency range of 0.1MHz to about 40MHz. This capacitor set-up consists of two horizontal-parallel plates with an adjustable separation for insertion of a dielectric specimen (PCC). While useful in research, this approach is not practical for field implementation In this research, a capacitance probe has been developed for field application. The probe consists of two planar conducting plates and is made of flexible materials for placement on exposed surfaces of the specimens to be tested. The calibration method of both capacitive systems has been extensively studied to minimize systematic errors in the measurement process. These two measurement systems will be discussed and compared to one another on the basis of sensitivity and measurement repeatability. / Master of Science

Le contrôle non destructif (CND) a pour but de détecter et de caractériser d’éventuels défauts présents dans des pièces mécaniques. Les techniques ultrasonores actuelles utilisent des capteurs multiéléments associés à des chaînes d’instrumentations et d’acquisitions de données multi capteurs en parallèles. Compte tenu de la masse de données à traiter, l’analyse de ces dernières est généralement effectuée hors ligne. Des travaux en cours, au Commissariat à l’Energie Atomique (CEA), consistent à développer et évaluer différentes méthodes d’imageries avancées, basées sur la focalisation synthétique. Les algorithmes de calculs induits nécessitent d’importantes opérations itératives sur un grand volume de données, issues d’acquisition multiéléments. Ceci implique des temps de calculs important, imposant un traitement en différé. Les contraintes industrielles de caractérisation de pièces mécaniques in situ imposent de réaliser la reconstruction d’images lors de la mesure et en temps réel. Ceci implique d’embarquer dans l’appareil de mesure, toute l’architecture de calcul sur les données acquises des capteurs. Le travail de thèse a donc consisté à étudier une famille d’algorithmes de focalisation synthétique pour une implantation temps réel sur un instrument de mesure permettant de réaliser l’acquisition de données. Nous avons également étudié une architecture dédiée à la reconstruction d’images par la méthode de Focalisation en Tout Point (FTP). Ce travail a été réalisé dans le cadre d’une collaboration avec l’équipe ACCIS de l’institut d’Electronique Fondamentale, Université de Paris Sud. Pour ce faire, notre démarche s’est inspirée de la thématique de recherche d’Adéquation Algorithme Architecture (A3). Notre méthodologie, est basée sur une approche expérimentale consistant dans un premier temps en une décomposition de l’algorithme étudié en un ensemble de blocs fonctionnels (calculs/transferts). Cela nous a permis de réaliser l’extraction des blocs pertinents de calculs à paralléliser et qui ont une incidence majeure sur les temps de traitement. Nous avons orienté notre stratégie de développement vers une conception flot de donnée. Ce type de modélisation permet de favoriser les flux de données et de réduire les flux de contrôles au sein de l’architecture matérielle. Cette dernière repose sur une plateforme multi-FPGA. La conception et l’évaluation de telles architectures ne peuvent se faire sans la mise en place d’outils logiciels d’aide à la validation tout au long du processus de la conception à l’implantation. Ces outils faisant partie intégrante de notre méthodologie. Les modèles architecturaux des briques de calculs ont été validés au niveau fonctionnel puis expérimental, grâce à la chaîne d’outils développée. Cela inclus un environnement de simulation nous permettant de valider sur tables les briques partielles de calculs ainsi que le contrôle associé. Enfin, cela a nécessité la conception d’outils de génération automatique de vecteurs de tests, à partir de données de synthèses (issues de l’outil simulation CIVA développé par le CEA) et de données expérimentales (à partir de l’appareil d’acquisition de la société M2M-NDT). Enfin, l’architecture développée au cours de ce travail de thèse permet la reconstruction d’images d’une résolution de 128x128 pixels, à plus de 10 images/sec. Ceci est suffisant pour le diagnostic de pièces mécaniques en temps réel. L’augmentation du nombre d’éléments capteurs ultrasonores (128 éléments) permet des configurations topologiques plus évoluées (sous forme d’une matrice 2D), ouvrant ainsi des perspectives vers la reconstruction 3D (d’un volume d’une pièce). Ce travail s’est soldé par une mise en œuvre validée sur l’instrument de mesure développé par la société M2M-NDT. / Non-destructive Evaluation (NDE) regroups a set of methods used to detect and characterize potential defects in mechanical parts. Current techniques uses ultrasonic phased array sensors associated with instrumentation channels and multi-sensor data acquisition in parallel. Given the amount of data to be processed, the analysis of the latter is usually done offline. Ongoing work at the French “Commissariat à l’Energie Atomique” (CEA), consist to develop and evaluate different methods of advanced imaging based on synthetic focusing. The Algorithms induced require extensive iterative operations on a large volume of data from phased array acquisition. This involves important time for calculations and implies offline processing. However, the industrial constraint requires performing image reconstruction in real time. This involves the implementation in the measuring device, the entire computing architecture on acquired sensor data. The thesis has been to study a synthetic focusing algorithm for a real-time implementation in a measuring instrument used to perform ultrasonic data acquisition. We especially studied an image reconstruction algorithm called Total Focusing Method (TFM). This work was conducted as part of collaboration with the French Institute of Fundamental Electronics Institute team of the University of Paris Sud. To do this, our approach is inspired by research theme called Algorithm Architecture Adequation (A3). Our methodology is based on an experimental approach in the first instance by a decomposition of the studied algorithm as a set of functional blocks. This allowed us to perform the extraction of the relevant blocks to parallelize computations that have a major impact on the processing time. We focused our development strategy to design a stream of data. This type of modeling can facilitate the flow of data and reduce the flow of control within the hardware architecture. This is based on a multi- FPGA platform. The design and evaluation of such architectures cannot be done without the introduction of software tools to aid in the validation throughout the process from design to implementation. These tools are an integral part of our methodology. Architectural models bricks calculations were validated functional and experimental level, thanks to the tool chain developed. This includes a simulation environment allows us to validate partial calculation blocks and the control associated. Finally, it required the design of tools for automatic generation of test vectors, from data summaries (from CIVA simulation tool developed by CEA) and experimental data (from the device to acquisition of M2M –NDT society). Finally, the architecture developed in this work allows the reconstruction of images with a resolution of 128x128 pixels at more than 10 frames / sec. This is sufficient for the diagnosis of mechanical parts in real time. The increase of ultrasonic sensor elements (128 elements) allows more advanced topological configurations (as a 2D matrix) and providing opportunities to 3D reconstruction (volume of a room). This work has resulted in implementation of validated measurement instrument developed by M2M -NDT.

The purpose of this thesis is non-destructive determination of residual stress state in the multi-pass welded steel plates by Magnetic Barkhausen Noise (MBN) technique. To control the effectiveness of the developed procedure, continuous MBN measurements on the heat affected zone and parent metal of the welded plates were performed. In the experimental part, various steel plates were welded with different number of weld passes. Various series of samples were prepared for residual stress and for angular deflection measurements. Microstructural investigation and hardness measurements were also conducted. The results were discussed to evaluate the effectiveness of MBN measurements to monitor the changes in the residual stress state in the welded components as a function of weld pass number.

The objective of this thesis is to design and implement a rapid, reconfigurable and portable reader for wirelessly interrogating inductively coupled passive sensors. While the current method of impedance analyzer is sensitive and an accurate, the instruments used are bulky and slow, substantially hampering in-field testing and interrogation of sensors. Current methods cannot provide a quantifiable measure on minimum necessary read-speeds and instrument accuracy desirable for rapid sensing applications. This work summarizes the design and hardware implementation of two reader methods that address the aforementioned requirements. Both reader methods are based on a reflectometer approach: Swept-frequency Reflectometer Reader and Switched-frequency Interrogation Technique (SWIFT). The first method is a much faster alternative to in-lab and in-field testing for structural health monitoring, and is intended as an immediate replacement for the impedance analyzer method. Switched-frequency Interrogation is specifically designed to satisfy the need for rapid and accurate reading, potentially for in-motion sensing applications. This method provides a way of empirically relating minimum necessary read-time required for desired read-ranges. It also facilitates quantification of uncertainty in measurements, which is very critical in determining instrument accuracy in-field. The system design and implementation of both methods are described in detail and experimental results are presented to benchmark the performance of the readers. Issues of instrument reliability and practical limitations are also discussed, with potential solutions. Both methods are intended as universal techniques for wirelessly interrogating coupled passive sensors, not limited to their current form of implementation. / text

For successful identification of structural defects in plates and pipes, it is essential to understand structural wave propagation characteristics such as dispersion relations. Analytical approaches to identify the dispersion relations of homogeneous, simple plates and circular pipes have been investigated by many researchers. However, for plates or pipes with irregular cross-sectional configurations or multi-layered composite structures, it is almost impossible to obtain the analytical dispersion relations and associated mode shapes. In addition, full numerical modeling approaches such as finite element (FE) methods are not economically feasible for high (e.g., ultrasonic) frequency analyses where an extremely large number of discretized meshes are required, resulting in significantly expensive computation. In order to address these limitations, Hybrid Analytical/Finite Element Methods (HAFEMs) are developed to model composite plates and pipes in a computationally-efficient manner. When a pipe system is used to transport a fluid, the dispersion curves obtained from a “hollow” pipe model can mislead non-destructive evaluation (NDE) results of the pipe system. In this study, the HAFEM procedure with solid elements is extended by developing fluid elements and solid-fluid boundary conditions, resulting in the dispersion curves of fluid-filled pipes. In addition, a HAFEM-based acoustic transfer function approach is suggested to consider a long pipe system assembled with multiple pipe sections with different cross-sections. For the validation of the proposed methods, experimental and full FE modeling results are compared to the results obtained from the HAFEM models. In order to detect structural defect locations in shell structures from defect-induced, subtle wave reflection signals and eliminate direct-excitation-induced and boundary-reflected, relatively-strong wave signals, a time-frequency MUSIC algorithm is applied to ultrasonic wave data measured by using an array of piezoelectric transducers. A normalized, structurally-damped, cylindrical 2-D steering vector is proposed to increase the spatial resolution of time-frequency MUSIC power results. A cross-shaped array is selected over a circular or linear array to further improve the spatial resolution and to avoid the mirrored virtual image effects of a linear array. Here, it is experimentally demonstrated that the proposed time-frequency MUSIC beamforming procedure can be used to identify structural defect locations on an aluminum plate by distinguishing the defect-induced waves from both the excitation-generated and boundary-reflected waves.

In this thesis, several aspects of GPR data processing for RC bridge decks are studied. First, autofocusing techniques are proposed to replace the previous expensive and unreliable human visual inspections during the iterative migration process for the estimation of the velocity/dielectric permittivity distribution from GPR data. Second, F-K filtering with dip relaxation is proposed for interference removal that is important for both imaging and the performance of post-processing techniques including autofocusing techniques and CS-based migration studied in this thesis. The targeted interferes here are direct waves and cross rebar reflections. The introduced dip relaxation is for accommodating surface roughness and medium inhomogeneity. Third, the newly developed CS-based migration is modified and evaluated on GPR data from RC bridge decks. A more accurate model by accounting for impulse waveform distortion that leads to less modeling errors is proposed. The impact of the selection of the regularization parameter on the comparative amplitude reservation and the imaging performance is also investigated, and an approach to preserve the comparative amplitude information while still maintaining a clear image is proposed. Moreover, the potential of initially sampling the time-spatial data with uniform sampling rates lower than that required by traditional migration methods is evaluated.