Dissertations / Theses on the topic 'Robust excitation'

Thesis (Ph. D.)--West Virginia University, 2001.
Title from document title page. Document formatted into pages; contains x, 121 p. : ill. (some col.). Vita. Includes abstract. Includes bibliographical references (p. 99-103).

Model Reference Adaptive Control (MRAC) is a widely studied adaptive control methodology that aims to ensure that a nonlinear plant with significant modeling uncertainty behaves like a chosen reference model. MRAC methods attempt to achieve this by representing the modeling uncertainty as a weighted combination of known nonlinear functions, and using a weight update law that ensures weights take on values such that the effect of the uncertainty is mitigated. If the adaptive weights do arrive at an ideal value that best represent the uncertainty, significant performance and robustness gains can be realized. However, most MRAC adaptive laws use only instantaneous data for adaptation and can only guarantee that the weights arrive at these ideal values if and only if the plant states are Persistently Exciting (PE). The condition on PE reference input is restrictive and often infeasible to implement or monitor online. Consequently, parameter convergence cannot be guaranteed in practice for many adaptive control applications. Hence it is often observed that traditional adaptive controllers do not exhibit long-term-learning and global uncertainty parametrization. That is, they exhibit little performance gain even when the system tracks a repeated command. This thesis presents a novel approach to adaptive control that relies on using current and recorded data concurrently for adaptation. The thesis shows that for a concurrent learning adaptive controller, a verifiable condition on the linear independence of the recorded data is sufficient to guarantee that weights arrive at their ideal values even when the system states are not PE. The thesis also shows that the same condition can guarantee exponential tracking error and weight error convergence to zero, thereby allowing the adaptive controller to recover the desired transient response and robustness properties of the chosen reference models and to exhibit long-term-learning. This condition is found to be less restrictive and easier to verify online than the condition on persistently exciting exogenous input required by traditional adaptive laws that use only instantaneous data for adaptation. The concept is explored for several adaptive control architectures, including neuro-adaptive flight control, where a neural network is used as the adaptive element. The performance gains are justified theoretically using Lyapunov based arguments, and demonstrated experimentally through flight-testing on Unmanned Aerial Systems.

Le présent sujet contribue à la modélisation et à la commande de systèmes de conversion éoliens basés sur un Générateur Synchrone à Double Excitation (GSDE). Le degré de liberté apporté par l’excitation bobinée offre la possibilité de travailler sur l’amélioration de l’efficacité énergétique sur les cycles de fonctionnement avec sollicitations aléatoires, comme dans le cas de l’éolien.On vise, à travers cette recherche, la mise en place de techniques de commande robuste d’un GSDE en vue de l’optimisation de son rendement aérodynamique et de la réduction de la fatigue mécanique. Dans ce contexte, un générateur hybride relié à une charge isolée pour des applications éoliennes est présenté. Des modèles linéaires sont tout d’abord établis. Ces modèles sont ensuite utilisés pour mettre en place les structures de contrôle appropriées tant du point de vue électrique que mécanique. Parallèlement, des modèles non linéaires très complets sont développés permettant une validation en simulation très poussée prenant en compte les harmoniques d’espace du générateur, les effets de commutation des convertisseurs et les effets de torsion sur l’arbre. En outre, un émulateur éolien de 3 kW est construit, en vue d'évaluer expérimentalement l’apport de notre générateur hybride dans le domaine éolien puis d'améliorer les contrôleurs synthétisés. Dans ce travail, deux stratégies de contrôle robuste pour une machine hybride sont implémentées et une comparaison entre un contrôleur CRONE et un contrôleur H∞ est établie. Des résultats très satisfaisants sont obtenus avec une meilleure performance du CRONE par rapport au H∞. Outre la problématique de l’optimisation de la production du système de conversion éolien, on a aussi cherché à réduire le taux d’harmonique en recourant à deux solutions : le filtrage passif et la réduction des ondulations du couple électromagnétique par action sur le courant d’excitation. Bien qu’on ait apporté des améliorations pour de grandes vitesses de rotation de la génératrice, ces solutions restent insuffisantes pour une connexion de l’architecture proposée au réseau électrique...Une fois l’applicabilité de la GSDE dans le domaine éolien prouvée pour le cas d’un système éolien de 3 kW, nous sommes passé à une puissance plus réaliste en interfaçant notre modèle électrique avancé avec un modèle aéroélastique, disponible sur le logiciel FAST. Ce dernier permet de prendre en compte les éléments mécaniques, les couplages et les éventuelles flexibilités. La turbine choisie pour l’étude est la turbine WindPACT de puissance 1.5MW. Dans cette partie, des commandes robustes traitant la réduction de la fatigue mécanique sont élaborées.Le modèle de la turbine WindPACT basé sur le générateur hybride est finalement connecté au réseau, les lois de commande nécessaires pour cette connexion sont implémentées puis validées sous la plateforme Matlab Simulink
This subject contributes to the modeling and control of a wind conversion system based on a Double Excitation Synchronous Generator (DESG). The degree of freedom provided by the wound excitation allows the improvement of the energy efficiency on the operation’s cycles with random solicitations, as it is the case for wind turbines.The aim of this research is to implement robust control techniques for the DESG in order to optimize its aerodynamic efficiency and to reduce its mechanical loads. In this context, a hybrid generator connected to an isolated load for wind applications is presented. First, linear models are established. These models are used to set up the appropriate control structures from both an electrical and mechanical point of view. At the same time, very complete nonlinear models are developed allowing a validation in an advanced simulation platform taking into account the space harmonics of the generator, the switching effects of the converters and the torsional effects on the shaft. Moreover, a 3 kW wind emulator is built to evaluate the contribution of the hybrid generator in the wind conversion systems field and then improve the synthesized controllers. Two robust control strategies for a hybrid machine are implemented and a comparison between a CRONE controller and a H∞ controller is presented. Satisfying results are obtained with a better performance for the CRONE regulator compared to H∞ one. In addition to the problem of optimizing the production of the wind energy conversion system, attempts have also been made to reduce the generator harmonic distortion ratios by using two solutions: passive filtering and reduction of the electromagnetic torque ripple by acting on the excitation current. Although there are improvements at high rotation speeds, these solutions are not sufficient for a connection of the proposed architecture to the grid.Once the applicability of the DESG in the wind energy field has been proven in the case of a 3 kW wind conversion system, we have considered a more realistic case. To this end, we have interfaced the developed advanced electric model with an aeroelastic model available on the FAST software, to take into account the mechanical couplings and the flexibilities. The turbine chosen for the study is the WindPACT 1.5MW turbine. In this part, robust controllers dealing with the reduction of mechanical fatigue are developed.The model of WindPACT turbine based on the hybrid generator is finally connected to the grid and the control laws necessary for this connection are implemented and validated under the Matlab Simulink platform

In recent years, linear prediction voice encoders have become very efficient in terms of computing execution time and channel bandwidth usage while providing, in the absence of im- pulsive noise, natural sounding synthetic speech signals. This good performance has been achieved via the use of a maximum likelihood parameter estimation of an auto-regressive model of order ten that best fits the speech signal under the assumption that the signal and the noise are Gaussian stochastic processes. However, this method breaks down in the presence of impulse noise, which is common in practice, resulting in harsh or non-intelligible audio signals. In this paper, we propose a robust estimator of correlation, the Phase-Phase correlator that is able to cope with impulsive noise. Utilizing this correlator, we develop a Robust Mixed Excitation Linear Prediction encoder that provides improved audio quality for voiced, unvoiced, and transition speech segments. This is achieved by applying a statistical test to robust Mahalanobis distances for identifying the outliers in the corrupted speech signal, which are then replaced with filtered signals. Simulation results reveal that the proposed method outperforms in variance, bias, and breakdown point three other robust approaches based on the arcsin law, the polarity coincidence correlator, and the median- of-ratio estimator without sacrificing the encoder bandwidth efficiency and the compression gain while remaining compatible with real-time applications. Furthermore, in the presence of impulsive noise, the proposed speech encoder speech perceptual quality also outperforms the state of the art in terms of mean opinion score.
Doctor of Philosophy
Impulsive noise is a natural phenomenon in everyday experience. Impulsive noise can be analogous to discontinuities or a drastic change in natural progressions of events. Specifically in this research the disrupting events can occur in signals such as speech, power transmission, stock market, communication systems, etc. Sudden power outage due to lighting, maintenance or other catastrophic events are some of the reasons why we may experience performance degradation in our electronic devices. Another example of impulsive noise is when we play an old damaged vinyl records, which results in annoying clicking sounds. At the time instance of each click, the true music or speech or simply the audible waveform is completely destroyed. Other examples of impulse noise is a sudden crash in the stock market; a sudden dive in the market can destroy the regression and future predictions. Unfortunately, in the presence of impulsive noise, classical methods methods are unable to filter out the impulse corruptions. The intended filtering objective of this dissertation is specific, but not limited, to speech signal processing. Specifically, research different filter model to determine the optimum method of eliminating impulsive noise in speech. Note, that the optimal filter model is different for time series signal model such as speech, stock market, power systems, etc. In our studies we have shown that our speech filter method outperforms the state of the art algorithms. Another major contribution of our research is in speech compression algorithm that is robust to impulse noise in speech. In digital signal processing, a compression method entails in representing the same signal with less data and yet convey the the same same message as the original signal. For example, human auditory system can produce sounds in the range of approximately 60 Hz and 3500 Hz, another word speech can occupy approximately 4000 Hz in frequency space. So the challenge is, can we compress speech in one of half of that space, or even less. This is a very attractive proposition because frequency space is limited but the wireless service providers desires to service as many users as possible without sacrificing quality and ultimately maximize the bottom line. Encoding impulse corrupted speech produces harsh quality of synthesized audio. We have shown if the encoding is done with the proposed method, synthesized audio quality is far superior to the sate of the art.

In recent years more emphasis has been placed on in-situ condition based monitoring of engineering systems and structures. Aerospace components are manufactured from composite materials more often. Structural health monitoring (SHM) systems are required in the aerospace industry to monitor the safety and integrity of the structure and will ensure that composites reach its full potential within the industry. Damage detection techniques form an integral part of such SHM systems. With this work a damage detection technique is developed for intended eventual use on composite structures, but starting first on isotropic structures. The damage mechanism that is of interest is delamination damage in composites. A simple numerical equivalent is implemented here however. Two damage indicators, the strain cumulative damage factor (SCDF) and the strain-frequency damage level (SFDL) are introduced. The respective damage indicators are calculated from output-only strain and acceleration response data. The effectiveness of the system to detect damage in the structure is critically evaluated and compared to other damage detection techniques such as the natural frequency method. The sensitivity to damage and performance of both these indicators is examined numerically by evaluating two deterministic damage cases. The numerical study is enhanced through the use of an updated finite element model. The minimum number of sensors capable of detecting the presence and locate damage spatially is determined from numerical simulations. Monte Carlo type analysis is performed by letting the damaged area vary stochastically and calculating the respective damage indicators. The model updating procedure from measured mobility frequency response functions (FRFs) is described. The application of the technique to real structures is examined experimentally. Two test structures with two different damage scenarios are examined. The spatial location and presence of damage can be established from both the SCDF and SFDL values, respectively. The spatial location obtained from the SCDF values corresponded to the known damage location for both the numerical and experimental study. The SFDL proved to be more sensitive than the natural frequency method and could be used to calculate the level of damage within the structure.
Dissertation (MEng)--University of Pretoria, 2009.
Mechanical and Aeronautical Engineering
unrestricted

Le problème posé et traité dans ce mémoire est celui de l'identification des paramètres des robots. Cette identification est indispensable pour la commande de robots basée sur le modèle de référence. La majeure difficulté de l'identification de ces paramètres provient de la dimension du système. En effet le nombre de conditionnement de la matrice d'observation non linéaire est à minimiser. Afin de remédier à cette difficulté, ce travail montre qu'il est possible de décomposer le problème et de procéder à une identification décentralisée. Le problème du calcul des trajectoires à excitation convenable peut être mieux résolu par cette décomposition. Ainsi il est montré que la décentralisation permet d'obtenir de meilleures conditions d'excitation. Les méthodes d'optimisation non linéaires sous contraintes sont utilisées pour le calcul des trajectoires. Lorsque ces méthodes sont appliquées au cas décentralisés, les calculs peuvent être notablement accélères. Les résultats de l'estimation décentralisée montrent que la rapidité de convergence des paramètres est nettement augmentée par rapport à l'estimation centralisée. Quant à la sensibilité des paramètres, une étude montre qu'il est possible de l'évaluer en fonction des perturbations sur les trajectoires

Dans cette thèse, des méthodes dites de Lyapunov sont proposées afin de résoudre des problèmes liés à la coordination distribuée des systèmes multiagent, plus précisément, un groupe de systèmes (agents) non-linéaires formés de robots mobiles non-holonomes est considéré. Pour ce groupe de systèmes, des lois de commande distribuée sont proposées dans le but de résoudre des problèmes de type leader-suiveur en formation et aussi des problèmes de type formation sans-leader par une approche de consensus, sous différentes hypothèses sur le graphe de communication et surtout sur les vitesses du leader.L'originalité de ce travail est dans l'approche proposée pour l'étude de stabilité de la boucle fermée, cette approche consiste à transformer les deux derniers problèmes en des problèmes de stabilisation globale asymptotique d'un ensemble invariant. L’analyse de stabilité est basée sur la construction de fonction de Lyapunov et de fonction de Lyapunov-Karasovskii strictes pour des classes de systèmes non-linéaires variant dans le temps présentant des retards bornés et variant dans le temps
In this thesis, we propose a Lyapunov based approaches to address some distributedsolutions to multi-agent coordination problems, more precisely, we consider a groupof agents modeled as nonholonomic mobile robots, we provide a distributed controllaws in order to solve the leader-follower and the leaderless consensus problems under different assumptions on the communication graph topology and on the leader’strajectories. The originality of this work relies on the closed-loop analysis approach, that is, it consists on transforming the last two problems into a global stabilization problem of an invariant set. The stability analysis is mainly based on the construction of strict Lyapunov functions and strict Lyapunov-Krasovskii functionals for a classes of nonlinear time-varying and/or delayed systems

Thesis (S.M.)--Massachusetts Institute of Technology, Dept. of Mechanical Engineering, 2009.
Includes bibliographical references (leaf 71).
A simple, compact, yet powerful robotic winch, called "Winch-Bot," is presented in this thesis. The Winch-Bot is an underactuated robot having only one controllable axis. Although hanging a load with merely one cable, it is capable of moving it in a large workspace by swinging the load dynamically based on parametric self-excitation. The generated trajectories can be used for a variety of tasks, from moving material to cyclic inspection of surfaces. The basic principle and design concept of the Winch-Bot are first described, followed by dynamic modeling and analysis. Two trajectory generation problems are solved. One is point-to-point transfer of a load, and the other is the tracking of a continuous path. It will be shown that the system can track a given geometric trajectory, although the tracking velocity cannot be determined arbitrarily due to the underactuated nature of dynamics. A prototype Winch-Bot is designed and built, and point-to-point, continuous path, and parametric excitation control are implemented.
by Daniel Philip Cunningham.
S.M.

In this masters thesis we consider a method for experimental identification of the inertial parameters of an industrial robot, using measured torques and joint angles. A dynamic model of the first three joints of the robot has been identified.

To achieve good identification results, it is important to carefully choose the trajectory for the experimental identification. A method to generate trajectories using two suggested design criteria has been used and evaluated using an ABB industrial robot, and one of them yields good identification results.

Denna rapport behandlar en metod för experimentell identifiering av en stelkroppsmodell för en industrirobot. Metoden använder sig av uppmätta moment och armvinklar för att identifiera parametrarna i en dynamikmodell för robotens tre huvudaxlar.

För att erhålla bra identifieringsresultat är det viktigt att välja en lämplig identifieringsbana och i detta arbete har en metod för generering av banor använts och utvärderats för två olika designkriterier. Experiment har utförts på en industrirobot från ABB och vi har erhållit goda identifieringsresultat med ett av designkriterierna.

Les multiples applications liées aux systèmes multi-agents en réseau, tels que les satellites en formation, les oscillateurs couplés, les véhicules aériens sans pilote, entre autres, ont été, sans aucun doute, une motivation majeure dans le développement de cette thèse, qui est consacrée à l’étude du consensus de systèmes dynamiques et à la commande en formation de robots mobiles non holonomes. Dans le contexte du consensus, nous étudions la topologie en anneau avec de liens de communication variant dans le temps. Notamment, la communication peut être perdue pendant de longs intervalles de temps. Nous donnons de conditions suffisantes pour le consensus qui restent simples à vérifier, par exemple, en utilisant le théorème du petite gain. En suite, nous abordons le problème de consensus en supposant que la topologie de communication est variable. Nous établissons que le consensus est atteint à condition qu’il existe toujours un chemin de communication du type « spanning-tree » pendant un temps de séjour minimal. L'analyse s'appuie sur la théorie de stabilité des systèmes variant dans le temps et les systèmes à commutation. Dans le contexte de la commande en formation de véhicules autonomes nous adressons le problème de commande en suivi de trajectoire sur ligne droite en suivant une approche type maître-esclave. Nous montrons que le suivi global peut être obtenu à partir d’un contrôleur qui possède la propriété d’excitation persistante. En gros, le mécanisme de stabilisation dépend de l’excitation du système par une quantité qui est proportionnelle à l’erreur de suivi. Ensuite, la méthode est utilisée pour résoudre le problème de suivi de formation de plusieurs véhicules interconnectés sur la base d’une topologie « spanning-tree ». Nous donnons des conditions de stabilité concernant les modèles cinématique et dynamique, en utilisant la seconde méthode de Lyapunov
The multiple applications related to networked multi-agent systems such as satellite formation flying, coupled oscillators, air traffic control, unmanned air vehicles, cooperative transport, among others, has been undoubtedly a watershed for the development of this thesis. The study of cooperative control of multi-agent systems is of great interest for his extensive field work and applications. This thesis is devoted to the study of consensus seeking of multi-agents systems and trajectory tracking of nonholonomic mobile robots.In the context of consensus seeking, first we study a ring topology of dynamic agents with time-dependent communication links which may disconnect for long intervals of time. Simple checkable conditions are obtained by using small-gain theorem to guarantee the achievement of consensus. Then, we deal with a network of dynamic agents with time-dependent communication links interconnected over a time-varying topology. We establish that consensus is reached provided that there always exists a « spanning-tree » for a minimal dwell-time by using stability theory of time-varying and switched systems. In the context of trajectory tracking, we investigate a simple leader-follower tracking controller for autonomous vehicles following straight lines. We show that global tracking may be achieved by a controller which has a property of persistency of excitation tailored for nonlinear systems. Roughly speaking the stabilisation mechanism relies on exciting the system by an amount that is proportional to the tracking error. Moreover, the method is used to solve the problem of formation tracking of multiple vehicles interconnected on the basis of a « spanning-tree » topology. We derive stability conditions for the kinematic and dynamic model by using a Lyapunov approach

by Chiu Kim Ming.
Thesis (M.Phil.)--Chinese University of Hong Kong, 1994.
Includes bibliographical references (leaves 103-108).
Chapter Chapter 1 --- Introduction --- p.1
Chapter 1.1 --- Speech production --- p.2
Chapter 1.2 --- Low bit rate speech coding --- p.4
Chapter Chapter 2 --- Speech analysis & synthesis --- p.8
Chapter 2.1 --- Linear prediction of speech signal --- p.8
Chapter 2.2 --- LPC vocoder --- p.11
Chapter 2.2.1 --- Pitch and voiced/unvoiced decision --- p.11
Chapter 2.2.2 --- Spectral envelope representation --- p.15
Chapter 2.3 --- Excitation --- p.16
Chapter 2.3.1 --- Regular pulse excitation and Multipulse excitation --- p.16
Chapter 2.3.2 --- Coded excitation and vector sum excitation --- p.19
Chapter 2.4 --- Multiband excitation --- p.22
Chapter 2.5 --- Multiband excitation vocoder --- p.25
Chapter Chapter 3 --- Dual-band and Quad-band excitation --- p.31
Chapter 3.1 --- Dual-band excitation --- p.31
Chapter 3.2 --- Quad-band excitation --- p.37
Chapter 3.3 --- Parameters determination --- p.41
Chapter 3.3.1 --- Pitch detection --- p.41
Chapter 3.3.2 --- Voiced/unvoiced pattern generation --- p.43
Chapter 3.4 --- Excitation generation --- p.47
Chapter Chapter 4 --- A low bit rate Quad-Band Excitation LSP Vocoder --- p.51
Chapter 4.1 --- Architecture of QBELSP vocoder --- p.51
Chapter 4.2 --- Coding of excitation parameters --- p.58
Chapter 4.2.1 --- Coding of pitch value --- p.58
Chapter 4.2.2 --- Coding of voiced/unvoiced pattern --- p.60
Chapter 4.3 --- Spectral envelope estimation and coding --- p.62
Chapter 4.3.1 --- Spectral envelope & the gain value --- p.62
Chapter 4.3.2 --- Line Spectral Pairs (LSP) --- p.63
Chapter 4.3.3 --- Coding of LSP frequencies --- p.68
Chapter 4.3.4 --- Coding of gain value --- p.77
Chapter Chapter 5 --- Performance evaluation --- p.80
Chapter 5.1 --- Spectral analysis --- p.80
Chapter 5.2 --- Subjective listening test --- p.93
Chapter 5.2.1 --- Mean Opinion Score (MOS) --- p.93
Chapter 5.2.2 --- Diagnostic Rhyme Test (DRT) --- p.96
Chapter Chapter 6 --- Conclusions and discussions --- p.99
References --- p.103
Appendix A Subroutine of pitch detection --- p.A-I - A-III
Appendix B Subroutine of voiced/unvoiced decision --- p.B-I - B-V
Appendix C Subroutine of LPC coefficients calculation using Durbin's recursive method --- p.C-I - C-II
Appendix D Subroutine of LSP calculation using Chebyshev Polynomials --- p.D-I - D-III
Appendix E Single syllable word pairs for Diagnostic Rhyme Test --- p.E-I

Mel-frequency cepstral coefficients (MFCCs) are widely adopted in speech recognition as well as speaker recognition applications. They are extracted to primarily characterize the spectral envelope of a quasi-stationary speech segment. It was shown that cepstral features are closely related to the linguistic content of speech. Besides the magnitude-based cepstral features, there are resources in speech, e.g, the phase and excitation source, are believed to contain useful properties for speaker discrimination. Moreover, in real situations, there are large variations exist between the development and application scenarios for a speaker recognition system. These include channel mismatch, recording apparatus mismatch, environmental variation, or even change of emotional/healthy state of speakers. As a consequence, the magnitude-based features are insufficient to provide satisfactory and robust speaker recognition accuracy. Therefore, the exploitation of complementary features with MFCCs may provide one solution to alleviate the deficiency, from a feature-based perspective.
Speaker recognition (SR) refers to the process of automatically determining or verifying the identity of a person based on his or her voice characteristics. In practical applications, a voice can be used as one of the modalities in a multimodal biometric system, or be the sole medium for identity authentication. The general area of speaker recognition encompasses two fundamental tasks: speaker identification and speaker verification.
Wang, Ning.
Adviser: Pak-Chung Ching.
Source: Dissertation Abstracts International, Volume: 73-06, Section: B, page: .
Thesis (Ph.D.)--Chinese University of Hong Kong, 2011.
Includes bibliographical references (leaves 177-193).
Electronic reproduction. Hong Kong : Chinese University of Hong Kong, [2012] System requirements: Adobe Acrobat Reader. Available via World Wide Web.
Electronic reproduction. [Ann Arbor, MI] : ProQuest Information and Learning, [201-] System requirements: Adobe Acrobat Reader. Available via World Wide Web.
Abstract also in Chinese.

碩士
國立臺灣大學
土木工程學研究所
105
In recent years, tuned mass dampers have been widely used to control the vibrations. TMDs are often adopted in high-rise buildings, long-span bridges, and machines of high-tech factory to reduce the vibration and the discomfort. However, the values of TMD parameters determine the effectiveness. There are several design formulas for the structure without inherent damping, but in reality the structure has inherent damping. Therefore, it is necessary to use the iteration method to obtain the optimal design variables, and this study integrates two algorithms to find the best solution and optimal design variables rapidly and efficiently. Generally, the tuned mass damper design emphasizes the reducing effect rather than the robustness. However, TMD is sensitive to the frequency between the structure and the TMD. The reducing effect deteriorates if the frequency shift away from the optimal value. Therefore, this study proposes the optimal design method of robust tuned mass damper, considering the condition of the reducing effect and the variability of the frequency simultaneously, and then make a comparison between the effectiveness of the general design and the robust design. In the researches, it is quite common to design TMDs by simplifying MDOF model to SDOF model rather than the real structure. If the analytical results are similar, it is much rapider and easier to solve the problem by using SDOF model. Furthermore, the performance of system with TMD sometimes becomes worse when the structure is affected by the earthquakes, so the actual earthquake records are used to examine the result statistically in this study, and finally make a comparison between the performance of the general design and the robust design under the earthquakes.

The human respiratory system plays a crucial role in breathing and swallow ing. However, it also plays an essential role in speech production, which is unique to humans. Speech production involves expelling air from the lungs. As the air flows from the lungs to the lips, some kinetic energy gets con verted to sound. Different structures modulate the generated sound, which is finally radiated out of the lips. The speech consists of various informa tion such as linguistic content, speaker identity, emotional state, accent, etc. Apart from speech, there are various scenarios where the sound is generated in the human respiratory system. These could be due to abnor malities in the muscles, motor control unit, or the lungs, which can directly affect generated speech as well. A variety of sounds are also generated by these structures while breathing including snoring, Stridor, Dysphagia, and Cough. The source filter (SF) model of speech is one of the earlier models of speech production. It assumes that speech is a result of filtering an excita tion or source signal by a linear filter. The source and filter are assumed to be independent. Even though the SF model represents the speech pro duction mechanism, there needs to be a tractable way of estimating the excitation and the filter. The estimation of both of them given speech falls under the general category of signal deconvolution problem, and, hence, there is no unique solution. There are several variations of the source-filter model in the literature by assuming different structures on the source/filter. There are various ways to estimate the parameters of the source and the filter. The estimated parameters are used in various speech applications such as automatic speech recognition, text to speech, speech enhancement etc. Even though the SF model is a model of speech production, it is used in applications including Parkinson’s Disease classification, asthma classification. The existing source filter models show much success in various appli cations, however, we believe that the models mainly lack two respects. The first limitation is that these models lack the connection to the physics of sound generation or propagation. The second limitation of the cur rent models is that they are not fully probabilistic. The inherent nature of the airflow is stochastic because of the presence of turbulence. Hence, probabilistic modeling is necessary to model the stochastic process. The probabilistic models come with several other advantages: 1) systematically inducing the prior knowledge into the models through probabilistic priors, 2) the estimation of the uncertainty of the model parameters, 3) allows sampling of new data points 4) evaluation of the likelihood of the observed speech. We start with the governing equation of sound generation and use a simplified geometry of the vocal folds. We show that the sound generated by the vocal folds consists of two parts. The first part is because of the difference between the subglottal and supra glottal pressure difference. The second part is because of the sound generated by turbulence. The first kind is dominant in the voiced sounds, and the second part is dominant in the unvoiced sounds. We further assume the plane wave propagation in the vo cal tract, and there is no feedback from the vocal tract on the vocal folds. The resulting model is the excitation passing through an all-pole filter, and the excitation is the sum of two signals. The first signal is quasi-periodic, and the shape of each cycle depends on the time-varying area of the glottis. The second part is stochastic because the turbulence is modeled as a white noise passed through a filter. We further convert the model into a proba bilistic one by assuming the following distribution on the excitations and filters. We model the excitation using a Bernoulli Gaussian distribution. Filter coefficients are modeled using the Gaussian distribution. The noise distribution is also Gaussian. Given these distributions, the likelihood of the speech can be derived as a closed-form expression. Similarly, we im pose an appropriate prior to the model’s parameters and make a maximum a posteriori (MAP) estimation of the parameters. The MAP estimation of parameters can be computationally complex. But the model assumption can be changed/approximated with respect to the application and result ing in different estimation procedures. To validate the model, we apply this model to seven applications as follows: 1. Analysis and Synthesis: This ap plication is to understand the representation power of the model. 2. Robust GCI detection: This shows the usefulness of estimated excitation, and the probabilistic modeling helps to incorporate the second-order statistics for robust the excitation estimation. 3. Probabilistic glottal inverse filtering: This application shows the usefulness of the prior distribution on filters. 4. Neural speech synthesis: We show that the model’s reformulation with the neural network results in a computationally efficient neural speech synthe sis. 5. Prosthetic esophageal (PE) to normal speech conversion: We use the probabilistic model for detecting the impulses in the noisy signal to convert the PE speech to normal speech. 6. Robust essential vocal tremor classification: The usefulness of robust excitation estimation in pathological speech such as essential vocal tremor. 7. Snorer group classification: Based on the analogy between voiced speech production and snore production, the derived model is applicable for snore signals. We also use the parameter of the model to classify the snorer groups.

Adaptive control of a nonlinear time varying (NTV) plant, such as a robotic manipulator, is intended to tolerate the unmodeled disturbances and the uncertain parameters of the dynamic model. Most of the previous research has been focused on NTV plants with bounded and "slowly-varying" plant terms. Almost all adaptive controllers require persistent excitations to guarantee stable tracking in the presence of unmodeled disturbances. The new adaptive controllers developed in this work provide stable and robust performance without persistent excitations and the "slowly-varying" assumption. Moreover, the uncertainties of a NTV plant model are not required to be bounded. This allows one to treat some potentially unbounded dynamics as disturbances. Stability and robustness analysis of adaptive controllers under the relaxed conditions is an essential part of this study. A major problem arising in robotic control is parameter uncertainty. The linear parameterization approach is also implemented in this work to deal with the parameter uncertainty. An innovative algorithm for determining the manipulator "regressor" (a coefficient matrix in parameter-linearized form of robot dynamics) is developed. Based on this algorithm a robust self-tuning controller is designed. The control law is proved to be robust with respect to parameter errors and disturbances. The robustness of the controller relaxes the requirement for the parameter estimator, and leads to a stable system without persistent excitations.
Graduate