Le contrôle de santé structural ou Structural Health Monitoring (SHM) des matériaux constitue une démarche fondamentale pour la maîtrise de la durabilité et de la fiabilité des structures en service. Au-delà des enjeux industriels et humains qui ne cessent de s’accroître en termes de sécurité et de fiabilité, le contrôle de santé doit faire face à des exigences de plus en plus élaborées. Les nouvelles stratégies de contrôle de santé doivent non seulement détecter et identifier l’endommagement mais aussi quantifier les différents phénomènes qui en sont responsables. Pour atteindre cet objectif, il est nécessaire d’accéder à une meilleure connaissance des processus d’endommagement. Par ailleurs, ceux-ci surviennent fréquemment sous l’effet de sollicitations mécaniques et environnementales. Ainsi, il est indispensable, d’une part, d’élaborer des méthodes de traitement des signaux permettant d’estimer les effets des conditions environnementales et opérationnelles, dans un contexte de l’analyse des événements précurseurs des mécanismes d’endommagement, et, d’autre part, de définir les descripteurs d’endommagement les plus adaptés à cette analyse. Cette étude propose donc des méthodes de traitement du signal permettant d’atteindre cet objectif, dans un premier temps, pour l’estimation des effets externes sur les ondes multidiffusées dans un contexte de contrôle de santé actif et, dans un second temps, pour l’extraction d’un indicateur d’endommagement à partir de l’analyse des signaux d’émission acoustique dans un contexte de contrôle de santé passif. Dans la première partie de ce travail, quatre méthodes de traitement du signal sont proposées. Celles-ci permettent de prendre en compte les variations des conditions environnementales dans la structure, qui dans le cadre de cette thèse, se sont limitées au cas particulier du changement de la température. En effet, les variations de température ont pour effet de modifier les propriétés mécaniques du matériau et par conséquent la vitesse de propagation des ondes ultrasonores. Ce phénomène entraîne alors une dilatation temporelle des signaux acoustiques qu’il convient d’estimer afin de suivre les variations de température. Quatre estimateurs de coefficients de dilatation sont alors étudiés : Il s’agit de l’intercorrélation à fenêtre glissante, utilisée comme méthode de référence, la méthode du stretching, l’estimateur à variance minimale et la transformée exponentielle. Les deux premières méthodes ont été déjà validées dans la littérature alors que les deux dernières ont été développées spécifiquement dans le cadre de cette étude. Par la suite, une évaluation statistique de la qualité des estimations est menée grâce à des simulations de Monte-Carlo utilisant des signaux de synthèse. Ces signaux sont basés sur un modèle de signal multidiffusé prenant en compte l’influence de la température. Une estimation sommaire de la complexité algorithmique des méthodes de traitement du signal complète également cette phase d’évaluation. Enfin, la validation expérimentale des méthodes d’estimation est réalisée sur deux types de matériaux : Tout d’abord, dans une plaque d’aluminium, milieu homogène dont les caractéristiques sont connues, puis, dans un second temps dans un milieu fortement hétérogène prenant la forme d’une plaque composite en verre/epoxy. Dans ces expériences, les plaques sont soumises à différentes températures dans un environnement thermique contrôlé. Les estimations de température sont alors confrontées à un modèle analytique décrivant le comportement du matériau. La seconde partie de ce travail concerne la caractérisation in situ des mécanismes d’endommagement par émission acoustique dans des matériaux hétérogènes. Les sources d’émission acoustique génèrent des signaux non stationnaires... / Structural health monitoring (SHM) of materials is a fundamental measure to master thedurability and the reliability of structures in service. Beyond the industrial and human issuesever increasing in terms of safety and reliability, health monitoring must cope with demandsincreasingly sophisticated. New health monitoring strategies must not only detect and identifydamage but also quantify the various phenomena involved in it. To achieve this objective, itis necessary to reach a better understanding of the damage process. Moreover, they frequentlyoccur as a result of mechanical and environmental stresses. Thus, it is essential, first, to developsignal processing methods for estimating the effects of environmental and operational conditions,in the context of the analysis of precursor events of damage mechanisms, and on theother hand, to define the damage descriptors that are the most suitable to this analysis. Thisstudy proposes signal processing methods to achieve this goal. At first, to the estimation ofexternal effects on the scattered waves in an active health control context, in a second step, tothe extraction of a damage indicator from the signals analysis of acoustic emission in a passivehealth monitoring context.In the first part of this work, four signal processing methods are proposed. These allow takinginto account the variation of environmental conditions in the structure, which in this thesis,were limited to the particular case of temperature change. Indeed, temperature changes have theeffect of altering the mechanical properties of the material and therefore the propagation velocityof ultrasonic waves. This phenomenon then causes a dilation of the acoustic signals that shouldbe estimated in order to monitor changes in temperature. Four estimators of dilation coefficientsare then studied: the intercorrelation sliding window, used as reference method, the stretchingmethod, the minimum variance estimator and the exponential transform. The first two methodshave already been validated in the literature while the latter two were developed specificallyin the context of this study. Thereafter, a statistical evaluation of the quality of estimates isconducted through Monte Carlo simulations using synthetic signals. These signals are basedon a scattered signal model taking into account the influence of temperature. A raw estimateof the computational complexity of signal processing methods also completes this evaluationphase. Finally, the experimental validation of estimation methods is performed on two types ofmaterial: First, in an aluminum plate, homogeneous medium whose characteristics are known,then, in a second step in a highly heterogeneous environment in the form of a compositeglass/epoxy plate. In these experiments, the plates are subjected to different temperatures in acontrolled thermal environment. The temperature estimates are then faced with an analyticalmodel describing the material behavior.The second part of this work concerns in situ characterization of damage mechanisms byacoustic emission in heterogeneous materials. Acoustic emission sources generate non-stationarysignals. The Hilbert-Huang transform is thus proposed for the discrimination of signals representativeof four typical sources of acoustic emission in composites: matrix cracking, debondingfiber/matrix, fiber breakage and delamination. A new time-frequency descriptor is then definedfrom the Hilbert-Huang transform and is introduced into an online classification algorithm. Amethod of unsupervised classification, based on the k-means method, is then used to discriminatethe sources of acoustic emission and the data segmentation quality is evaluated. Thesignals are recorded from blank samples, using piezoelectric sensors stuck to the surface of thematerial and sensitive samples (sensors integrated within the material)...

Human body temperature is an important indicator of physical performance and condition in terms of comfort, heat or cold stress. The aim of this research was to develop Temperature Sensing Fabric (TSF) for continuous temperature measurement in healthcare applications. The study covers the development and manufacture of TSF by embedding fine metallic wire into the structure of textile material using a commercial computerised knitting machine. The operational principle of TSF is based on the inherent propensity of a metal wire to respond to changes in temperature with variation in its electrical resistance. Over 60 TSF samples were developed with combinations of different sensing elements, two inlay densities and highly textured polyester yarn as the base material. TSF samples were created using either bare or insulated wires with a range of diameters from 50 to 150 μm and metal wires of nickel, copper, tungsten, and nickel coated copper. In order to investigate the Temperature-Resistance (T-R) relationship of TSF samples for calibration purposes, a customised test rig was developed and monitoring software was created in the LabVIEW environment, to record the temperature and resistance signals simultaneously. TSF samples were tested in various thermal environments, under laboratory conditions and in practical wear trials, to analyse the relationship between the temperature and resistance of the sensing fabric and to develop base line specifications such as sensitivity, resistance ratio, precision, nominal resistance, and response time; the influence of external parameters such as humidity and strain were also monitored. The regression uncertainty was found to be less than in ±0.1°C; the repeatability uncertainty was found to be less than ±0.5°C; the manufacturing uncertainty in terms of nominal resistance was found to be ± 2% from its mean. The experimental T-R relationship of TSF was validated by modelling in the thermo-electrical domain in both steady and transient states. A maximum error of 0.2°C was found between the experimental and modelled T-R relationships. TSF samples made with bare wire sensing elements showed slight variations in their resistance during strain tests, however, samples made with insulated sensing elements did not demonstrate any detectable strain-dependent-resistance error. The overall thermal response of TSF was found to be affected by basal fabric thickness and mass; the effect of RH was not found to be significant. TSF samples with higher-resistance sensing elements performed better than lower-resistance types. Furthermore, TSF samples made using insulated wire were more straightforward to manufacture because of their increased tensile strength and exhibited better sensing performance than samples made with bare wire. In all the human body wear trials, under steady-state and dynamic conditions both sensors followed the same trends and exhibited similar movement artifacts. When layers of clothing were worn over the sensors, the difference between the response of the TSF and a high-precision reference temperature were reduced by the improved isothermal conditions near the measurement site.

Reducing unscheduled maintenance is important for aircraft operators. There are significant costs if flights must be delayed or cancelled, for example, if spares are not available and have to be shipped across the world. This thesis describes three methods of aircraft health condition monitoring and prediction; one for system monitoring, one for forecasting and one combining the two other methods for a complete monitoring and prediction process. Together, the three methods allow organizations to forecast possible failures. The first two use decision trees for decision-making and genetic optimization to improve the performance of the decision trees and to reduce the need for human interaction. Decision trees have several advantages: the generated code is quickly and easily processed, it can be altered by human experts without much work, it is readable by humans, and it requires few resources for learning and evaluation. The readability and the ability to modify the results are especially important; special knowledge can be gained and errors produced by the automated code generation can be removed. A large number of data sets is needed for meaningful predictions. This thesis uses two data sources: first, data from existing aircraft sensors, and second, sound and vibration data from additionally installed sensors. It draws on methods from the field of big data and machine learning to analyse and prepare the data sets for the prediction process.

The poly(glycidyl methacrylate)-modified magnetite nanoparticles, Fe3O4-PGMA NPs, were investigated and applied in nondestructive flaw detection of polymeric materials in this research. The Fe3O4 endowed magnetic property to the materials for flaw detection while the PGMA promoted colloidal stability and prevented particle aggregation. The magnetite nanoparticles (Fe3O4 NPs) were successfully synthesized by coprecipitation and then surface-modified with PGMA to form PGMA-modified Fe3O4 NPs by both grafting-from and grafting-to approaches. For the grafting-from approach, the Fe3O4 NPs were surface-functionalized with α-bromo isobutyryl bromide (BIBB) to form BIB-modified Fe3O4 NPs (Fe3O4-BIB NPs) with covalent linkage. The resultant Fe3O4-BIB NPs were used as surface-initiators to grow PGMA by surface-initiated atom transfer radical polymerization (SI-ATRP). For the grafting-to approach, the Fe3O4 NP were surface-functionalized with (3-mercaptopropyl)triethoxysilane (MCTES) to form MCTES-modified Fe3O4 NPs (Fe3O4-MCTES NPs). The PGMA with Br-end group was pre-synthesized by ATRP and then was grafted to the surface of the Fe3O4-MCTES NPs by coupling reaction. Both bare and modified Fe3O4 NPs exhibited superparamagnetism and the existence of iron oxide in the form of Fe3O4 was confirmed. The particle size of individual Fe3O4 NPs was about 8 – 24 nm but they aggregated to form clusters. The PGMA-modified NPs formed stable dispersion in chloroform and had larger cluster sizes than the unmodified ones because of the PGMA polymer layer. However, the uniformity of the NP clusters could be improved with PGMA surface grafting. The PGMA surface layer of the grafting-from (Fe3O4-gf-PGMA) NPs was thin and dense while that of the grafting-to (Fe3O4-gt-PGMA) NPs was thick and loose. The hydrodynamic diameters (Zave) of Fe3O4-gf-PGMA NP clusters could be controlled between 176 to 643 nm, dependent on the PGMA contents and reaction conditions. During SI-ATRP, side reactions happened and caused NP aggregation as well as increase of size of NP clusters. However, the aggregation has been minimized through optimization of reaction conditions. Oppositely, Zave values of Fe3O4-gt-PGMA NPs had little variation of about 120 – 190 nm. And the PGMA content of the Fe3O4-gt-PGMA NPs was limited to 12.5% because of the spatial hindrance during grafting process. The saturation magnetization (Ms) of the unmodified Fe3O4 NPs was about 77 emu/g, while those of the grafting-from and grafting-to Fe3O4-PGMA NPs were 50 – 66 emu/g and 63 – 70 emu/g, respectively. For Fe3O4-PGMA NPs with similar Fe3O4 contents, the grafting-to NPs had slightly higher Ms than the grafting-from counterparts. In addition, the Ms of both kinds of the Fe3O4-PGMA NPs with higher Fe3O4 content (> 87%) were also higher than that of the fluidMAG-Amine, the commercially available amine-modified MNPs. Besides, both kinds of Fe3O4-PGMA NPs also had much higher Fe3O4 contents and Ms values than most of the reported PGMA-modified MNPs. The magnetic epoxy nanocomposites (MENCs) were prepared by blending the modified Fe3O4 NPs into bisphenol A diglycidyl ether (BADGE)-based epoxy system and the distributions of both kinds of the PGMA-modified NPs were much better than that of the oleic acid-modified Fe3O4 NPs. Similar to the NPs, the MENCs also exhibited superparamagnetism. By cross-section TEM observation, the grafting-to Fe3O4-PGMA NPs formed more homogeneous distributions with smaller cluster size than the grafting-from counterparts and gave higher Ms of the MENCs. Nondestructive flaw detection of surface and sub-surface defects could be successfully achieved by brightness contrast of images given through eddy current testing (ET) method, which is firstly reported. The mechanical properties of the materials were influenced very slightly when 2.5% or lower Fe3O4-gt-PGMA NPs were present while the presence of the Fe3O4-gf-PGMA NPs (1 – 2.5 %) gave mild improvement of the storage modulus and increase of the glass-rubber transition temperature(Tg) of the MENCs. Furthermore, the Fe3O4-PGMA NPs could be evenly coated onto the functionalized ultra-high molecular weight poly(ethylene) (UHMWPE) textiles. The Fe3O4-gt-PGMA NPs were coated on the textile in order to prepare NP-coated textile-reinforced composite. Preliminary result of ET measurement showed that the Fe3O4-gt-PGMA NPs coated on the textiles could visualize the structure of the textile hidden inside and their relative depth. Accordingly, the incorporation of MNPs to polymers opens a new pathway of damage-free structural health monitoring of polymeric materials.:1. Introduction 2. Theoretical section 2.1. Magnetite Nanoparticles (MNPs) 2.2. Applications of MNPs 2.3. Atom transfer radical polymerization (ATRP) 2.4. Magnetic nanocomposites (MNCs) 2.5. Damage-free structural health monitoring (SHM) using MNPs 3. Objective of the work 4. Materials, methods and characterization 4.1. Materials 4.2. Methods 4.3. Formation of polymeric magnetic nanocomposites 4.4. Characterization 5. Results and discussions 5.1. Unmodified magnetite nanoparticles (Fe3O4 NPs) 5.2. Oleic acid-modified (Fe3O4–OA) NPs 5.3. PGMA-modified NPs by grafting-from approach (Fe3O4-gf-PGMA NPs) 5.4. PGMA-modified NP by grafting-to approach (Fe3O4-gt-PGMA NPs) 5.5. Comparison between grafting-from and grafting-to Fe3O4-PGMA NPs 5.6. Magnetic epoxy nanocomposites (MENCs) 5.7. Fiber-reinforced epoxy nanocomposites 6. Conclusions and outlook 7. Appendix 8. List of figures, schemes and tables 9. References Versicherung Erklaerung List of publications

Among various nanofillers, carbon nanotubes (CNTs) have attracted a significant attention due to their excellent physical properties. Incorporation of a very low amount of CNTs in polymer matrices enhances mechanical, thermal and optical properties of conductive polymer nanocomposites (CPNs) tremendously. For mechanical sensors, the piezoresistive property of CNTs/polymer nanocomposites exhibits a great potential for the realization of stable, sensitive, tunable and cost-effective strain sensors. Achieving homogeneous CNTs dispersion within the polymer matrices, understanding their complex piezoresistivity and conduction mechanisms, as well as the response of the nanocomposites under humidity and temperature effects, is highly required for the realization of piezoresistive CNTs/polymer based nanocomposites. This research primarily aims to synthesize and characterize CNTs/polymer based strain sensitive nanocomposites, which are cost-effective, applicable on both rigid and flexible substrates and require a non-complex fabrication process. A comprehensive understanding of the complex conduction and piezoresistive mechanisms of CNTs/polymer nanocomposites and their responses under humidity and temperature effects is another purpose of this thesis. For this purpose, synthesis and complex electromechanical characterization of multiwalled carbon nanotubes (MWCNTs)/epoxy nanocomposites are realized. In order to realize strain sensors for the strain range up to 1 % the use of epoxy is focused due to its good adhesion, dimensional stability, and good mechanical properties. The nanocomposites with up to 1 wt.% MWCNTs are synthesized by a non-complex direct mixing method and the final nanocomposites are deposited on flexible Kapton and rigid FR4 substrates and their corresponding morphological, electrical, electromechanical, as well as the response of the nanocomposite under humidity and temperature influences, are examined. The deformation over the sensor area is tested by digital image correlation (DIC) under quasi-static uniaxial tension. Quantitative piezoresistive characterization is performed by electrochemical impedance spectroscopy (EIS) over a wide range of frequencies. Further, dispersion quality of MWCNTs in the epoxy polymer matrix is monitored by scanning electron microscopy (SEM). Additionally, in order to tailor the piezoresistivity of the strain sensor, an R-C equivalent circuit is derived based on the impedance responses and the corresponding parameters are extracted from the applied strain. Obtained SEM images confirm that MWCNTs/epoxy nanocomposites with different MWCNTs concentrations have a good homogeneity and dispersion. Atomic force microscopy (AFM) analysis show that the samples have relatively good surface topography and fairly homogeneous CNTs networks. Higher sensitivity is achieved in particular at the concentrations close to the percolation threshold. A non-linear piezoresistive behavior is observed at low MWCNTs concentrations due to the dominance of tunneling effect. The strain sensitive nanocomposites deposited on FR4 substrates present high-performance strain sensing properties, including high sensitivity, good stability, and durability after cyclic loading and unloading. In addition, MWCNTs/epoxy nanocomposites show quite a small creep, low hysteresis under cyclic tensile and compressive loadings and fast response and recovery times. Nanocomposites provide an opportunity to measure 2-D strain in one position including amplitude and direction for complex configuration of structures in real-time systems or products. In contrast to present solutions for multi-directional strain sensing, MWCNTs/epoxy based nanocomposites give promising results in terms of durability, easy-processability, and tunable piezoresistivity. Unlike commercially-available approaches for crack/damage identification, MWCNTs/epoxy nanocomposites are capable of detecting the applied crack directly over a certain area. From the humidity influence, it has been found that resistance of nanocomposites increases with the increase of humidity exposure due to swelling of the polymer. Temperature investigations show that MWCNTs/epoxy nanocomposites give negative temperature coefficient (NTC) response due to thermal activation of charge carriers and the temperature sensitivity increases with the increase of filler concentration. The proposed approach can be further developed by combining differently fabricated sensors for realizing a compact structural health monitoring system or multi-functional sensor, where pressure, strain, temperature, and humidity can be monitored simultaneously. / Unter den verschiedenen Nanofillern haben CNTs aufgrund ihrer hervorragenden physikalischen Eigenschaften eine bedeutende Aufmerksamkeit erregt. Die Einarbeitung einer sehr geringen Menge an CNTs in Polymermatrizen verbessert die mechanischen, thermischen und optischen Eigenschaften von CPNs enorm. Für mechanische Sensoren bietet die piezoresistive Eigenschaft von CNTs/Polymer-Nanokompositen ein großes Potenzial zur Realisierung stabiler, empfindlicher, abstimmbarer und kostengünstiger Dehnungssensoren. Die Erzielung einer homogenen CNT-Dispersion innerhalb der Polymermatrizen, das Verständnis ihrer komplexen Piezoresistivitäts- und Leitungsmechanismen sowie die Reaktion der Nanokomposite unter Feuchte- und Temperatureinflüssen ist für die Realisierung piezoresistiver CNTs/Polymer-basierter Nanokomposite unerlässlich. Diese Arbeit zielt darauf ab, CNTs/polymerbasierte dehnungsempfindliche Nanokomposite herzustellen und zu charakterisieren. Diese Nanokompositen sollen kostengünstig, sowohl auf starren als auch auf flexiblen Substraten anwendbar sein und ein nicht komplexes Herstellungsverfahren erfordern. Ein umfassendes Verständnis der komplexen leitungs- und piezoresistive Mechanismen von CNTs/ Polymer-Nanokompositen und deren Reaktionen unter Feuchtigkeits- und Temperatureinflüssen ist ein weiteres Ziel dieser Arbeit. Zu diesem Zweck werden Synthese und komplexe elektromechanische Charakterisierung von MWCNTs/epoxy nanocomposites realisiert. Um Dehnungssensoren für den Dehnungsbereich bis zu 1 % realisieren zu können, wird der Einsatz von Epoxy aufgrund seiner guten Haftung, Dimensionsstabilität und guten mechanischen Eigenschaften fokussiert. Zufällig verteilte MWCNTs mit bis zu 1 wt.% MWCNTs-Konzentration ist durch ein direktes Mischen synthetisiert und die Nanokomposite werden auf flexiblen Kapton und starren FR4 Substraten durch Siebdruck appliziert und anschließend deren morphologische, elektrische, elektromechanische sowie die Reaktion des Nanocomposits unter Feuchtigkeits- und Temperatureinflüssen untersucht. Die Verformung über den Sensorbereich wird duch die Digital Image Correlation (DIC) Methode unter quasi-statischer uniaxialer Spannung getestet. Die quantitative piezoresistive Charakterisierung wird mit elektrische Impedanzspektroskopie (EIS) in einem breitem Frquenzspektrum durchgeführt. Ferner wird die Dispersionsqualität von MWCNTs in der Epoxidepolymermatrix durch Scanning Electron Microscopy (SEM) überprüft. Zusätzlich ist, um die Piezoresistivität des Dehnungssensors abzustimmen, eine RC-Äquivalenzschaltung auf der Grundlage der Impedanzantworten abgeleitet und die entsprechenden Parameter unter Belastung extrahiert. Erhaltene SEM-Bilder bestätigen, dass MWCNTs/Epoxide-Nanokomposite mit unterschiedlichen MWCNTs-Konzentrationen eine gute Homogenität und Dispersion aufweisen. Die atomic force microscopy (AFM) Untersuchung zeigt, dass die Proben relativ gute Oberflächentopographie und ziemlich homogene CNT-Netzwerke aufweisen. Eine höhere Empfindlichkeit wird insbesondere bei den Konzentrationen nahe der Perkolationsschwelle erreicht. Eine nichtlineare Piezoresistivität wird bei niedrigen MWCNTs Konzentrationen aufgrund der Dominanz des Tunnelwirkungseffekts beobachtet. Die auf FR4-Substraten applizierten dehnungsempfindlichen Nanokomposite weisen ausgezeichnete Dehnungsmessungseigenschaften einschließlich hohe Empfindlichkeit, gute Stabilität und Haltbarkeit nach zyklischer Be- und Entlastung auf. Darüber hinaus zeigen MWCNTs/Epoxide-Nanokomposite ein geringes Kriechen, eine kleine Hysterese unter zyklischen Zug- und Druckbelastungen, sowie schnelle Reaktionsund Wiederherstellungszeiten. Nanokomposite bieten die Möglichkeit, 2-D-Dehnungen in einer Position einschließlich Amplitude und Richtung innerhalb einer Materialstruktur in Echtzeitsystemen oder Produkten zu messen. Im Gegensatz zu aktuellen Lösungen für die multi-direktionale Dehnungsmessung, bieten die MWCNTs/Epoxide-Nanokomposite vielversprechende Ergebnisse in Bezug auf Langlebigkeit, leichte Verarbeitung und einstellbare Piezoresistivität. Im Unterschied zu kommerziell verfügbaren Ansätzen wird festgestellt, dassMWCNTs/Epoxide-Nanokomposite zur Riss-/Schadenserkennung in der Lage sind, den angelegten Riss direkt über einen bestimmten Bereich zu detektieren. Aus dem Einfluss der Feuchtigkeit hat sich herausgestellt, dass die Resistenz von Nanokompositen mit zunehmender Feuchtigkeitsbelastung durch Quellung des Polymers zunimmt. Temperaturuntersuchungen zeigen, dass MWCNTs/Epoxide-Nanokomposite aufgrund der thermischen Aktivierung von Ladungsträgern auf Temperatureinflüsse reagieren und die Temperaturempfindlichkeit mit der Erhöhung der Füllstoffkonzentration zunimmt. Der vorgeschlagene Ansatz kann durch die Kombination unterschiedlich hergestellte Sensoren zur Realisierung eines kompakten zur Überwachung des Zustands von Strukturen oder von multifunktionalen Sensoren weiterentwickelt werden, bei denen gleichzeitig Druck, Dehnung, Temperatur und Feuchtigkeit überwacht werden können.

The Silicon Carbide Bipolar Transistor is a device that is barely brought into real application so far. It features very low conduction losses and a high power density. The application is in some points different and unusual in comparison to the mainstream power semiconductors as IGBTs or MOSFETs. The Silicon Carbide Bipolar Transistor, the SiC-BJT, is a current driven device and the effort in driving is uncommonly high. As an outcome of the present work it can be said that it is more like a shift of requirements from the power semiconductor power unit to the driver stage. With consideration of all system losses, including driving losses, the final unoptimized COSIVU prototype inverter system gained an increase of efficiency of 40-60% in comparison to the IGBT-based reference system dependent on the applied load points. In terms of reliability and possible failure modes, the SiC-BJT behaves differently from the mainstream devices. One result of the project is that the chips itself are quite robust but the packaging needs some improvements. Thermal impedance spectroscopy is a method for detecting possible deterioration in the cooling path of a device. A method for temperature estimation of the SiC-BJT during on-state will be presented in this work. The electronic hardware for thermal impedance spectroscopy has been developed to do the measurements in a non-laboratory setup in the inverter in real application. Furthermore, the hardware implementation was realized on a very small space for integration into an in-wheel motor inverter system. / Der Siliciumkarbid Bipolartransistor ist ein leistungselektronisches Bauelement, was bis heute kaum über Labor- und Forschungsprojekte hinaus anwendungsnah zum Einsatz kam. Er verfügt über sehr geringe Durchlassverluste und eine hohe Leistungsdichte. Seine Verwendung und Anwendung ist in mancher Hinsicht anders und unüblich im Vergleich zu den etablierten leistungselektronischen Bauelementen wie IGBT und MOSFET. Der Siliciumkarbid Bipolartransistor, also der SiC-BJT, ist ein stromgesteuertes Bauteil, weswegen der Aufwand für die Treiber sehr hoch ist. Die praktische Arbeit im Rahmen des Forschungsprojektes „COSIVU“ mit den SiC-BJTs in Verbindung mit dem fertigen integrierten Invertersystem hat unter anderem gezeigt, dass es mehr eine Verschiebung der Anforderungen von der Leistungselektronik hin zu den Treibern für die Leistungselektronik ist. Unter Betrachtung der Verluste des gesamten Systems, einschließlich der Motor-, Treiber- und Steuerverluste, hat das fertige Prototyp-Invertersystem, welches durchaus noch Potential zur Optimierung besaß, eine deutliche Verbesserung des Wirkungsgrades erreicht. Gegenüber dem auf IGBT basierenden Referenz-Invertersystem, hat das COSIVU Invertersystem eine Verbesserung des Wirkungsgrades um 40-60 % erreicht. Eine Erkenntnis aus dem Forschungsprojekt in Bezug auf Zuverlässigkeit und mögliche Fehler und Defekte ist, dass der Chip selbst zwar ziemlich robust ist, aber dass die Gehäuse-, Aufbau- und Verbindungstechnik angepasst und verbessert werden sollte. Thermische Impedanzspektroskopie ist eine Methode um Verschlechterungen im Kühlpfad eines leistungselektronischen Halbleiters zu erkennen, was ein Kriterium für die Alterung des Bauteils ist. Eine Methode zur Bestimmung der Sperrschichttemperatur von SiC-BJTs während des normalen Durchlassbetriebes wird in dieser Arbeit vorgestellt. Die Platine für die thermische Impedanzspektroskopie wurde entwickelt, um die Messung in einem laborfernen Aufbau in einer echten Inverteranwendung durchzuführen. Zudem wurden die Platinenaufbauten auf sehr kleiner Fläche realisiert. Die Integration musste nämlich sehr kompakt gestaltet werden, da es sich um ein „in-wheel“ Motor-Inverter-System handelt, was zum größten Teil innerhalb eines Fahrzeugrades untergebracht ist.

<p>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.</p><br>

Aujourd’hui, trois technologies distinctes et complémentaires sont disponibles pour réaliser des mesures réparties de température, de déformation ou de vibration grâce à l’analyses des rétrodiffusion Raman, Brillouin et Rayleigh. Les besoins industriels actuels se portent sur la mesure répartie de déformation pour des infrastructures avec de longs linéaires, comme les canalisations, pour lesquelles une cartographie linéaire et en temps réel de leur état est demandée. Nous nous focalisons alors sur la conception d’un système de mesure Brillouin capable de mesurer de manière répartie et dynamique les déformations subies par une fibre optique. La méthode employée sera celle du flanc de frange ; elle a déjà été développée et expérimentée sur une architecture opto-électronique de type analyseur Brillouin (Brillouin-OTDA), nécessitant l’accès aux deux extrémités de la fibre optique. Dans notre cas, elle est implémentée sur une architecture fonctionnant en réflectométrie. Les résultats expérimentaux obtenus seront caractérisés et validés par la simulation des mesures de la déformation et du déplacement d’une canalisation supportée entre deux appuis simples ; un modèle mécanique, adapté à cette configuration et transposable sur des projets réels, est développé. Par le biais de partenaire industriels de Cementys, ce modèle est utilisé dans deux projets de surveillance de canalisation d’hydrocarbures dont les moyens d’installation et la finalité sont différents. / Today, three distinct and complementary technologies are available for distributed temperature, strain or vibration measurements with the analysis of Raman, Brillouin and Rayleigh backscattered light. Current industrial needs are distributed strain measurements for linear infrastructures, such as pipelines, for which linear and real-time strain distribution is required. The research work aims to design a new distributed and dynamic strain measurement system based on the analysis of spontaneous Brillouin backscatter by reflectometry. Slope assisted technique is used to accelerate the measurement acquisition, currently limited to static events because of their actual principle of sweep frequency acquisition of the Brillouin backscattering spectrum. The experimental results are characterized and validated by the simulation of the measurements of the deformation and displacement of a pipe supported between two simple supports. A mechanical model, adapted to this configuration and transposable on real projects, is developed. Through Cementys industrial partner, this model is then used for two monitoring project of pipelines with different installation facilities and purpose.

Dans une aviation où la sécurité des vols est au cœur des préoccupations des constructeurs, le contrôle de santé des structures est l'un des nouveaux pôles majeurs de recherche et développement engagé par la communauté aéronautique depuis ces dix dernières années. Un système SHM (structural Heath monitoring) intégré aux structures avioniques (tels que le sont déjà les systèmes de monitoring des moteurs) permettrait de : - rendre l’aviation plus sûre et éviterait certains des accidents aériens ; - réduire les coûts de maintenance ; - alléger, à terme, le poids total car cela permettrait de d’éviter les sur-renforcements structuraux actuels. Le travail développé durant cette thèse, dans le cadre d'un projet industriel, concerne le développement de solutions exploitant l'utilisation de nœuds piezoélectriques au sein de microsystèmes reconfigurables dédiés à la détection de défauts dans des éléments de structure d'avion. L'exploitation de données issues de la génération/capture d'ondes de Lamb ainsi que des techniques se basant sur l'étude de l'impédance électromécanique du capteur ont été développées et étudiées sur différents types de défauts identifiés tels que cracks, corrosion, délaminages etc... La méthode proposée repose sur la comparaison et l'évolution dans le temps de signatures de réseaux de capteurs utilisant l’effet piezoélectrique et placés sur des éléments choisis de structures avions. L'interface capteur-matériau a été spécialement étudiée afin de garantir le couplage le plus efficace possible. Les techniques de « monitoring » ainsi développées ont été testées sur des structures aéronautiques métalliques et des structures en matériaux composites simples/sandwichs extraites d’avions Airbus et ATR. Différentes solutions d’intégration de ces capteurs et nœuds ont été passées en revue et une démarche a été proposée, allant de l’architecture des effecteurs au conditionnement et à la transmission des signaux et informations d’intéret. Une nouvelle vision de l’électronique de détection de défauts, permettant de développer une instrumentation « universelle » de capteurs à travers une combinaison de circuits numériques/analogiques reconfigurables à entrées/sorties versatiles, a été implémentée et testée avec succès / Structural health monitoring (SHM) is certainly one of the key technologies required to provide the safety and the reliability of future aviation. Based on non-destructive testing, current on the ground periodical structural integrity inspections showed their limit as evidenced by the Columbia tragedy. For the time being, structural health monitoring technology has reached a good technology readiness level (TRL). However, the integration of these solutions into future aerospace vehicle will require advanced and innovative system architecture. Further, improved SHM techniques and alleged assessment algorithm will be necessary to ensure an embedded integration, as well as to fully exploit their sensing capability. For now, most of high critical embedded systems are based on federate architectures, where each calculator is dedicated to a specific function and to a unique kind of sensor. By consequence, the integration on the field of conventional SHM solutions is highly difficult due to the scale and the weight of the global electronics systems. Based on a fully reconfigurable micro-system, I propose in this thesis, a novel SHM approach that combines into a unique System on Chip: • Sensors instrumentation and interfacing using reconfigurable analog circuits• Signal management and conditioning using reconfigurable digital electronics • Heath diagnostic assessment algorithms using an embedded CPUBased on elastic guided waves and electromechanical impedance analysis, the presented solution is capable through piezoelectric sensors to detect different kinds of abnormal events such as impacts. Moreover, using advanced wavelet transform and signature comparison algorithms, the system is also capable to detect mechanical damages such as corrosion, cracks or delaminations ; no matter if the probed structure is in simple composite, honeycomb composite or metallic alloy. The feasibility was proven using multiples specimens directly extracted from Airbus and ATR airplanes. To cover large areas, the system is fully scalable and accepts a hardware upgrade through multiple communication ports and protocols. Moreover, the versatility of inputs/outputs interface allows the exploitation of multiple sensors in order to locate and triangulate flaws

Strukturüberwachung und Condition Monitoring spielen in vielen Gebieten der Technik eine große Rolle. Zur Überwachung von Leichtbaustrukturen aus faserverstärkten Kunststoffen bietet sich hierfür besonders die Körperschall-Analyse an. Am Markt etabliert sind hierfür piezoelektrische Signalaufnehmer. Diese Arbeit stellt eine kostengünstige Alternative in Form von mikromechanischen Körperschall-Sensoren vor. Eine Besonderheit stellt hierbei das Prinzip des mechanischen Bandpasses dar. Es wird die Elektronik- und Gehäuseentwicklung sowie die experimentelle Untersuchung dargelegt. / Structural health monitoring is of vital importance in many technical fields. For monitoring of lightweight structures made from fiber reinforced plastics especially acoustic emission testing is used. Commercially available transducers utilize the piezoelectric effect. This thesis introduces a cost efficient alternative in form of micromechanical sensors, in particular sensors using the principle of a mechanical bandpass. The design of electronics and the packaging as well as experimental investigations are provided.