Dissertations / Theses on the topic 'Signal processing; Spalling damage'

The objective of this thesis is to formulate and solve the sensor placement problem for damage localization in a sensor network. A Bayesian estimation problem is formulated with the time-of-flight (ToF) measurements. In this model, ToF of lamb waves, which are generated and received by piezoelectric sensors, is the total time for each wave to be transmitted, reflected by the target, and received by the sensor. The ToF of the scattered lamb wave has characteristic information about the target location. By using the measurement model and prior information, the target location is estimated in a centralized sensor network with a Monte Carlo approach. Then we derive the Bayesian Fisher information matrix (B-FIM) and based on that posterior Cramer-Rao lower bound (PCRLB), which sets a limit on the mean squared error (MSE) of any Bayesian estimator. In addition, we develop an optimal sensor placement approach to achieve more accurate damage localization, which is based on minimizing the PCRLB. Simulation results show that the optimal sensor placement solutions lead to much lower estimation errors than some sub-optimal sensor placement solutions.

The operation of our society depends heavily on infrastructure systems. To prevent failures and to reduce costs of maintenance, structural health monitoring (SHM) systems have been implemented on an increasing number of infrastructure systems. SHM systems have the potential to give reliable prediction of structural deterioration with less human safety risk and labor costs, and without interruption of normal operations. In the field of SHM, many techniques have been proposed in recent decades. Among these techniques, ultrasonic testing has been widely used for damage characterization in structures and materials. However, there remain many challenges in real-world SHM applications. For example, temperature variations can cause a significant decrease in performance of ultrasonic testing. Although there exist some temperature compensation techniques to improve the performance of ultrasonic testing under temperature variations, these techniques have their own limitations. This dissertation will focus on novel ultrasonic signal processing techniques for damage detection, quantification and temperature compensation. In Chapter 2, I will propose a modified optimal signal stretching (OSS) method and an singular value decomposition (SVD) method to solve the temperature compensation problem, where the OSS method (in its original form) failed to perform well for damage detection. In Chapter 3, I will study the statistical orthogonal relationship between temperature-induced and damage-induced ultrasonic change signals. The orthogonal relationship can be used to explain why SVD performs well under varying temperature conditions and why it also has the potential (under some conditions) to be directly used for damage detection and quantification. In Chapter 4, I viii will study the ultrasonic time-of-flight diffraction technique, which is used to quantify wall thickness loss of thick-walled aluminum tubes, because the conventional pulse-echo method did not perform well in my target application. In Chapter 5, I will propose a novel ultrasonic passband technique to quantify the alkali-silica reaction (ASR) caused cracking damage in concrete structures. This technique is based on the ultrasonic wave filtering effects of cracks in concrete. With the progress of ASR caused cracking damage in concrete, more high frequency components of ultrasonic waves are filtered out than low frequency components. The research work in this dissertation has the potential to help advance ultrasonic SHM techniques, to improve the real-world performance of ultrasonic SHM, to prevent failures of infrastructure systems, and to reduce the costs of maintenance if the proposed ultrasonic techniques can be implemented in real infrastructure systems in the future. However, some future work still needs to be done in order to implement the techniques studied in this dissertation in real-world applications.

Although nondestructive evaluation techniques have been implemented in many industry fields and proved to be useful, they are generally expensive, time consuming, and the results may not always be reliable. To overcome these drawbacks, structural health monitoring (SHM) systems has received significant attention in the past two decades. As structural systems are becoming more complicated and new materials are being developed, new methodologies, theories, and approaches in SHM have been developed for damage detection, diagnosis, and prognosis. Among the methods developed, the guided Lamb wave based SHM can be a promising technique for damage evaluation since it provides reliable damage information through signals propagating over large distance with little loss of amplitude. While this method is effective for damage assessment, the guided Lamb wave contains complicated mode characteristics, i.e. an infinite number of wave modes exist and these modes are generally dispersive. For this reason, a minimum number of wave modes and various signal processing algorithms are implemented to obtain better signal interpretations. Phased array beamsteering is an effective means for damage detection in guided Lamb wave SHM systems. Using this method, the wave energy can be focused at localized directions or areas by controlled excitation time delay of each array element. In this research, two types of transducers are utilized as phased array elements to compare beamsteering characteristics. Monolithic piezoceramic (PZT) transducers are investigated for beamsteering by assuming omnidirectional point sources for each actuator. MacroFiber Composite (MFC) transducers with anisotropic actuation are also studied, considering the wave main lobe width, main lobe magnitude, and side lobe levels. Analysis results demonstrate that the MFC phased arrays perform better than the PZT phased arrays for a range of beamsteering angles and have reduced main lobe width and side lobe levels. Experiments using the PZT and MFC phased arrays on an aluminum plate are also performed and compared to the analysis results. A time-frequency signal processing algorithm coupled with a machine learning method can form a robust damage diagnostic system. Four types of such algorithms, i.e. short time Fourier transform, Wigner-Ville distribution, wavelet transform, and matching pursuit, are investigated to select an appropriate algorithm for damage classification, and a spectrogram based on short time Fourier transform is adopted for its suitability. A machine learning algorithm called Adaboost is chosen due to its effectiveness and high accuracy performance. The classification is preformed using spectrograms and Adaboost for crack and corrosion damages. Artificial cracks and corrosions are created in Abaqus® to obtain the training samples consist of spectrograms. Several beam experiments in laboratory and additional simulations are also performed to get the testing samples for Adaboost. The analysis results show that not only correct damage classification is possible, but the confidence levels of each sample are acquired.
Ph. D.

L'obiettivo della disciplina nota con l'acronimo SHM (Structural Health Monitoring) è quello di incrementare la sicurezza di infrastrutture critiche, monitorandone in tempo reale l'integrità attraverso l'implementazione di sistemi sensoriali embedded. In questo contesto il presente lavoro di tesi riguarda, in particolare, lo sviluppo di due reti di sensori accelerometrici microelettromeccanici per l'analisi della dinamica di strutture attraverso l'estrazione di parametri modali, quali frequenze naturali di vibrazione e forme modali. Tecniche di signal processing sono state sviluppate per calcolare la densità spettrale di potenza (PSD) dei dati di accelerazione acquisiti, sulla base di approcci parametrici e non parametrici. Algoritmi nel dominio del tempo (TDD) e della frequenza (FDD), insieme all'identificazione a sorgenti indipendenti del secondo ordine (SOBI), sono stati implementati per la ricostruzione delle curve modali. L'affidabilità delle architetture studiate è stata valutata prima di tutto in condizioni operative nominali, quindi simulando sperimentalmente situazioni di anomalia di varia entità, così come impostando differenti schemi di acquisizione hardware e software. Gli esiti dell'elaborazione hanno mostrato in entrambi i casi un'alta e soddisfacente corrispondenza tra il modello analitico di riferimento e la risposta degli algoritmi implementati. Versatilità, facilità di riconfigurazione, scalabilità e compatibilità con installazioni a lungo termine sono alcuni fra i più importanti vantaggi delle circuiterie presentate. Il peso leggero e i costi ridotti, unitamente alla robustezza delle tecniche di elaborazione dei dati registrati, potenziano inoltre le capacità del sistema di fornire informazioni in tempo reale.

Thesis (M. S.)--Civil and Environmental Engineering, Georgia Institute of Technology, 2006.
Kim, Jin-Yeon, Committee Member ; Qu, Jianmin, Committee Member ; Jacobs, Laurence, Committee Chair. Includes bibliographical references.

Research in damage detection and structural health monitoring in engineering systems during their service life has received increasing attention because of its importance and benefits in maintenance and rehabilitation of structure. Though the concept of vibration-based damage detection has been in existence for decades, and several procedures have been proposed to date, its practical applications remain limited, considering the increased utilization of sensors to measure structural response at multiple points. In this thesis, use of acceleration response of the structure as a method of global damage detection is explored using instantaneous frequency and stiffness degradation methods. Instantaneous frequency was estimated using continuous wavelet transform of measured acceleration response of the structure subjected to ground motion. Complex Morlet Wavelet was used in the time-frequency analysis due to its ability to provide sufficient resolution in both time and frequency domains. This ability is important in analyzing nonstationary signals like earthquake response of structure containing sharp changes in the signal. The second method, called the stiffness degradation analysis, is based on estimating the time-varying stiffness. This estimation is done by fitting a moving least-square line to the force-displacement loop for the duration of the ground motion.A four-story shear building is used as the model structure for numerical analysis. Two damage scenarios are considered: single damage instant and multiple damage instants. Both scenarios assume that the damage occurs at a single location. In the numerical simulations, damage was modeled as a reduction in the stiffness of the first floor, and accelerations were computed at floor levels using state-space model. The two methods were compared in terms of their damage detection ability and it was shown that both methods can be used in detecting damage and the time at which the damage occurs. These methods can later be extended by simultaneously considering the correlations of responses at all floor levels. This extension may enable locating the damage and quantifying the severity of the damage.

Mechanical durability of the structures should be continuously monitored during their operation. Structural health monitoring (SHM) techniques are typically used for gathering the information which can be used for evaluating the current condition of a structure regarding the existence, location, and severity of the damage. Damage can occur in a structure after long-term operating under service loads or due to incidents. By detection of these defects at the early stages of their growth and nucleation, it would be possible to not only improve the safety of the structure but also reduce the operating costs. The main goal of this dissertation is to develop a reliable and cost-effective SHM system for inspection of composite and metallic structures. The Surface Response to Excitation (SuRE) method is one of the SHM approaches that was developed at the FIU mechatronics lab as an alternative for the electromechanical impedance method to reduce the cost and size of the equipment. In this study, firstly, the performance of the SuRE method was evaluated when the conventional piezoelectric elements and scanning laser vibrometer were used as the contact and non-contact sensors, respectively, for monitoring the presence of loads on the surface. Then, the application of the SuRE method for the characterization vii of the milling operation for identical aluminum plates was investigated. Also, in order to eliminate the need for a priori knowledge of the characteristics of the structure, some advanced signal processing techniques were introduced. In the next step, the heterodyne method was proposed, as a nonlinear baseline free, SHM approach for identification of the debonded region and evaluation of the strength of composite bonds. Finally, the experimental results for both methods were validated via a finite element software. The experimental results for both SuRE and heterodyning method showed that these methods can be considered as promising linear and nonlinear SHM approaches for monitoring the health of composite and metallic structures. In addition, by validating the experimental results using FEM, the path for further improvement of these methods in future researches was paved.

This PhD thesis deals with the experimental study of more precise rolling contact fatigue damage detection using acoustic emission method. A series of experiments was carried out on two representatives bearing steels and the analysis of sensitivity for the presence of contact damage was performed on selected parameters of acoustic emission. The extent of damage was classified into four classes and signal parameters the most characterizing the development of damage were correlated with the extent of damage. It was also verified the influence of lubricants on acoustic emission signals. The results have an impact on the implementation of more precise rolling contact fatigue tests and evaluation of parameters of acoustic emission signal. On the basis of experiments was established methodology for more precise RCF testing method using acoustic emission on test-rig AXMAT II.

The development cost of any civil infrastructure is very high; during its life span, the civil structure undergoes a lot of physical loads and environmental effects which damage the structure. Failing to identify this damage at an early stage may result in severe property loss and may become a potential threat to people and the environment. Thus, there is a need to develop effective damage detection techniques to ensure the safety and integrity of the structure. One of the Structural Health Monitoring methods to evaluate a structure is by using statistical analysis. In this study, a civil structure measuring 8 feet in length, 3 feet in diameter, embedded with thermocouple sensors at 4 different levels is analyzed under controlled and variable conditions. With the help of statistical analysis, possible damage to the structure was analyzed. The analysis could detect the structural defects at various levels of the structure.

The objective of this thesis is to develop vibration-based fault detection strategies for on-line condition monitoring of gear transmission systems. The study divides the thesis into three sections. First of all, the local stresses created by a root fatigue crack on a pinion spur gear are analyzed using a quasi-static finite element model and non-linear contact mechanics simulation. Backlash between gear teeth which is essential to provide better lubrication on tooth surfaces and to eliminate interference is included as a defect and a necessary part of transmission design. The second section is dedicated to fixed axis power trains. Torsional vibration is shown to cause teeth separation and double-sided impacts in unloaded and lightly loaded gearing drives. The transient and steady-state dynamic loading on teeth within a two stage crank-slider mechanism arising from backlash and geometric manufacturing errors is investigated by utilizing a non-linear multi-body dynamics software model. The multi-body model drastically reduces the computation time required by finite element methods to simulate realistic operation. The gears are considered rigid with elastic contact surfaces defined by a penalty based non-linear contact formulation. The third section examines a practical differential planetary transmission which combines two inputs and one output. Planetary gears with only backlash errors are compared to those containing both backlash and tooth defects under different kinematic and loading conditions. Fast Fourier Transform (FFT) analysis shows the appearance of side band modulations and harmonics of the gear mesh frequency. A joint time-frequency analysis (JTFA) during start-up reveals the unique vibration patterns for fixed axis gear train and differential planetary gear, respectively, when the contact forces increase during acceleration.

Dans le cadre d’une maintenance préventive de ses moteurs, Safran Aircraft Engines souhaite compléter ses opérations de diagnostic par un pronostic fiable de la durée de vie résiduelle des roulements. Suite à une agression, il y a actuellement une grande incertitude sur la durée de vie restante avant défaillance du roulement à partir du seuil d’observabilité vibratoire de l’endommagement. Les algorithmes actuels diagnostiquent un stade de dégradation approximatif et génèrent des messages d’alarme de différents niveaux, chaque niveau correspondant à un stade de dégradation différent, mêlant confiance et sévérité du diagnostic. Un aspect important du pronostic est la prise en compte des paramètres contextuels influant sur la vitesse de dégradation. Les objectifs de cette thèse sont de disposer de méthodes et d’outils permettant de quantifier un temps de fonctionnement restant avant défaillance de roulement en regard : - de la gravité de l’endommagement détecté, - des conditions environnementales de fonctionnement, - de la profondeur de pronostic souhaitée, Les contraintes industrielles associées à ces objectifs sont les suivantes : 1) Le pronostic devra être basé, a minima, sur des mesures vibratoires hautes fréquences de quelques kHz (accéléromètres ou microphones), des données contextuelles (les régimes de rotation des différents rotors, par exemple, ou encore les amplitudes des niveaux pilotés sur les régimes de rotation, révélateurs d’un chargement des paliers) 2) .Constituer une base de données d’essais issus d’un plan d’expériences : ces essais devront tenir compte des contraintes liées à la maîtrise des paramètres jugés significativement influents 3) Cette base de données devra prendre en compte la représentativité de l’environnement vibratoire d’un moteur d’avion. 4) Proposer un outil ou une méthode de pronostic en tenant compte de la nature du roulement à considérer
As part of preventive maintenance of its engines, Safran Aircraft Engines wishes to complete its diagnostic operations with a reliable prognosis of the residual life of the bearings. Following an attack, there is currently a great deal of uncertainty about the remaining life before bearing failure from the threshold of vibrational observability of the damage. Current algorithms diagnose an approximate stage of degradation and generate alarm messages of different levels, each level corresponding to a different stage of degradation, combining confidence and severity of diagnosis. An important aspect of the prognosis is the taking into account of the contextual parameters influencing the rate of degradation. The objectives of this thesis are to have methods and tools to quantify a running time remaining before bearing failure with regard to: - the severity of the damage detected, - the environmental conditions of operation, - the depth The industrial constraints associated with these objectives are as follows: 1) The prognosis should be based, at least, on high-frequency vibratory measurements of a few kHz (accelerometers), contextual data (the rotational speeds of the different rotors, for example, or the amplitudes of the levels piloted on them), rotation regimes, revealing a loading of the bearings) 2). Constituing a database of tests resulting from a plan of experiments: these tests will have to take into account the constraints related to the control of the parameters considered to be significantly influential 3) This database must take into account the representativity of the vibratory environment of an aircraft engine. 4) Propose a tool or method of prognosis taking into account the nature of the bearing to consider

Master of Science
Department of Civil Engineering
Hani G. Melhem
This study reports the results of an analytical, experimental and a numerical study (proof of concept study) on a proposed method for extracting the pseudo-free-vibration response of a structure using ambient vibration, usually of a random nature, as a source of excitation to detect any change in the dynamic properties of a structure that may be caused by damage. The structural response contains not only a random component but also a component reflecting the dynamic properties of the structure, comparable to the free vibration for a given initial condition. Structural response to the arbitrary excitation is recorded by one or several accelerometers with a desired data-collection frequency and resolution. The free-vibration response of the structure is then extracted from this data by removing the random component of the response by the method proposed in this study. The features of the free-vibration response of the structure extracted by a suitable method, namely Fast Fourier Transform (FFT) in this study, can be used for change detection. Possible change of the pattern of these features is dominantly linked to the change in dynamic properties of the system, caused by possible damage. To show the applicability of the concept, besides an analytical verification using Newmark’s linear acceleration method, two steel portal frames with different flexural stiffness were made in the steel workshop of the structural laboratory for an experimental study. These structures were also numerically modeled using a finite element software. A wireless accelerometer with a sampling frequency rate of 2046 Hz was affixed on the top of the physical structure, at the same location where the acceleration was recorded for the corresponding numerical model. The physical structure was excited manually by an arbitrary hit and the response of the structure to this excitation, in terms of the acceleration on the top of the structure, was recorded. The pseudo-free-vibration response was extracted and transferred into frequency domain using FFT. The frequency with the largest magnitude which is the fundamental frequency of the structure was traced. This was repeated for several independent excitations and the fundamental frequencies were observed to be the same, showing that the process can correctly identify the natural frequencies of the structure. Similarly, the numerical model was excited and for several base excitation cases, the fundamental frequencies were found to be the same. Considering the acceptable accuracy of the results from the two numerical models in simulating the response of their corresponding physical models, additional numerical models were analyzed to show the consistency and applicability of the proposed method for a range of flexural stiffness and damping ratio. The results confirm that the proposed method can precisely extract the pseudo-free-vibration response of the structures and detect the structural frequencies regardless of the excitation. The fundamental frequency is tied to the stiffness and a larger stiffness leads to a higher frequency, as expected, regardless of the simulated ambient excitation.

Recent advances in sensor technologies and data acquisition platforms have led to the era of Big Data. The rapid growth of artificial intelligence (AI), computing power and machine learning (ML) algorithms allow Big Data to be processed within affordable time constraints. This opens abundant opportunities to develop novel and efficient approaches to enhance the sustainability and resilience of Smart Cities. This work, by starting with a review of the state-of-the-art data fusion and ML techniques, focuses on the development of advanced solutions to structural health monitoring (SHM) and metamaterial design and discovery strategies. A deep convolutional neural network (CNN) based approach that is more robust against noisy data is proposed to perform structural response estimation and system identification. To efficiently detect surface defects using mobile devices with limited training data, an approach that incorporates network pruning into transfer learning is introduced for crack and corrosion detection. For metamaterial design, a reinforcement learning (RL) and a neural network based approach are proposed to reduce the computation efforts for the design of periodic and non-periodic metamaterials, respectively. Lastly, a physics-constrained deep auto-encoder (DAE) based approach is proposed to design the geometry of wave scatterers that satisfy user-defined downstream acoustic 2D wave fields. The robustness of the proposed approaches as well as their limitations are demonstrated and discussed through experimental data or/and numerical simulations. A roadmap for future works that may benefit the SHM and material design research communities is presented at the end of this dissertation.