This dissertation addresses novel techniques in determining gyroless micro-satellite attitude and attitude rate. The main objective of this thesis is to explore the possibility of using commercially available low cost micro-light star sensor as a stand-alone sensor for micro-satellite attitude as well as attitude rate determination. The objective is achieved by developing accurate and computationally efficient algorithms for the realization of onboard operation of a low fidelity star sensor. All the algorithms developed here are tested with the measurement noise presented in the catalog of the sensor array STAR-1000. A novel accurate second order sliding mode observer (SOSMO) is designed for discrete time uncertain linear multi-output system. Our design procedure is effective for both matched and unmatched bounded uncertain ties and/or disturbances. The bound on uncertainties and/or disturbances is assumed to be unknown. This problem is addressed in this work using the second order multiple sliding modes approach. Second order sliding manifold and corresponding sliding condition for discrete time system is defined similar on the lines of continuous counterpart. Our design is not restricted to a particular class of uncertain (matched) discrete time system. Moreover, it can handle multiple outputs unlike single out-put systems. The observer design is achieved by driving the state observation error and its first order finite difference to the vicinity of the equilibrium point (0,0) in a finite steps and maintaining them in the neighborhood thereafter. The estimation synthesis is based on Quasi Sliding Mode (QSM) design. The problem of designing sliding mode observer for a linear system subjected to unknown inputs requires observer matching condition. This condition is needed to ensure that the state estimation error is a asymptotically stable and is independent of the unknown input during the sliding motion. In the absence of a matching condition, asymptotic stability of the reduced order error dynamics on the sliding surface is not guaranteed. However, unknown bounded inputs guarantee bounded error on state estimation. The QSM design guarantees an ultimate error bound by incorporating Boundary Layer (BL) in its design procedure. The observer achieves one order of magnitude improvement in estimation accuracy than the conventional sliding mode observer (SMO) design for an unknown input. The observer estimation errors, satisfying the given stability conditions, converge to an ultimate finite bound (with in the specified BL) of O(T2), where T Is the sampling period. A relation between sliding mode gain and boundary layer is established for the existence of second order discrete sliding motion. The robustness of the proposed observer with respect to measurement noise is also analyzed. The design algorithm is very simple to apply and is implemented for two examples with different classes of disturbances (matched and unmatched) to show the effectiveness of the design. Simulation results show the robustness with respect to the measurement noise for SOSMO. Second order sliding mode observer gain can be calculated off-line and the same gain can work for large band of disturbance as long as the disturbance acting on the continuous time system is bounded and smooth. The SOSMO is simpler to implement on board compared to the other traditional nonlinear filters like Pseudo-Linear-Kalman-filter(PLKF); Extended Kalman Filter(EKF). Moreover, SMO possesses an automatic adaptation property same as optimal state estimator(like Kalman filter) with respect to the intensity of the measurement noise. The SMO rejects the noisy measurements automatically, in response to the increased noise intensity. The dynamic performance of the observer on the sliding surface can be altered and no knowledge of noise statistics is required. It is shown that the SOSMO performs more accurately than the PLKF in application to micro-satellite angular rate estimation since PLKF is not an optimal filter. A new method for estimation of satellite angular rates through derivative approach is proposed. The method is based on optic flow of star image patterns formed on a star sensor. The satellite angular rates are derived directly from the 2D-coordinates of star images. Our algorithm is computationally efficient and requires less memory allocation compared to the existing vector derivative approaches, where there is also no need for star identification. The angular rates are computed using least square solution method, based on the measurement equation obtained by optic flow of star images. These estimates are then fed into discrete time second order sliding mode observer (SOSMO). The performance of angular rate estimation by SOSMO is compared with the discrete time First order SMO and PLKF. The SOSMO gives the best estimates as compared to the other two schemes in estimating micro-satellite angular rates in all three axes. The improvement in accuracy is one order of magnitude (around1.7984 x 10−5 rad/ sec,8.9987 x 10−6 rad/ sec and1.4222 x 10−5 rad/ sec in three body axes respectively) in terms of standard deviation in steady state estimation error. A new method and algorithm is presented to determine star camera parameters along with satellite attitude with high precision even if these parameters change during long on-orbit operation. Star camera parameters and attitude need to be determined independent of each other as they both can change. An efficient, closed form solution method is developed to estimate star camera parameters (like focal length, principal point offset), lens distortions (like radial distortion) and attitude. The method is based on a two step procedure. In the first step, all parameters (except lens distortion) are estimated using a distortion free camera model. In the second step, lens distortion coefficient is estimated by linear least squares (LS) method. Here the derived camera parameters in first step are used in the camera model that incorporates distortion. However, this method requires identification of observed stars with the catalogue stars. But, on-orbit star identification is difficult as it utilizes the values of camera calibrating parameters that can change in orbit(detector and optical element alignment get change in orbit due to solar pressure or sudden temperature change) from the ground calibrated value. This difficulty is overcome by employing a camera self-calibration technique which only requires four observed stars in three consecutive image frames. Star camera parameters along with lens (radial and decentering) distortion coefficients are determined by camera self calibration technique. Finally Kalman filter is used to refine the estimated data obtained from the LS based method to improve the level of accuracy. We consider the true values of camera parameters as (u0,v0) = (512.75,511.25) pixel, f = 50.5mm; The ground calibrated values of those parameters are (u0,v0) =( 512,512) pixel, f = 50mm; Worst case radial distortion coefficient affecting the star camera lens is considered to be k1 =5 x 10−3 .Our proposed method of attitude determination achieves accuracy of the order of magnitude around 6.2288 x 10−5 rad,3.3712 x 10−5 radand5.8205 x 10−5 rad in attitude angles φ,θ and ψ. Attitude estimation by existing methods in the literature diverges from the true value since they utilize the ground calibrated values of camera parameters instead of true values. To summarize, we developed a formal theory of discrete time Second Order Sliding Mode Observer for uncertain multi-output system. Our methods achieve the desired accuracy while estimating satellite attitude and attitude rate using low fidelity star sensor data. Our methods require lower on-board processing requirement and less memory allocation; thus are suitable for micro-satellite applications. Thus, the objective of using low fidelity star sensor as stand-alone sensor in micro-satellite application is achieved.

Cette thèse s’intéresse au diagnostic et à l’observation de systèmes linéaires à commutations et à l’application au convertisseur multicellulaire série. L’objectif est de proposer des solutions pour des sous-systèmes non-observables au sens classique et dont des fautes continues ou discrètes peuvent être présentes. Après la présentation d’un état de l’art sur les techniques d’observation et de diagnostic pourles systèmes à commutations, le mémoire est scindé en deux parties. La première partie propose, d’une part, une stratégie d’estimation des états discret et continu d’un système linéaire à commutation soumis à une entrée inconnue. Un observateur hybride basé sur la théorie des modes glissants d’ordre supérieur est développé. D’autre part, deux procédures de diagnostic sont présentées. La première combine un observateur hybride et un diagnostiqueur pour détecter une faute continue. Pour la seconde, un diagnostic actif est défini sur la base de la théorie du test afin de détecter et d’isoler une faute discrète. Dans la seconde partie de ce mémoire, les étapes de la réalisation d’un convertisseur multicellulaire sont détaillées. Ensuite, un chapitre est dédié à la validation des approches théoriques d’observation et de diagnostic sur le convertisseur à trois cellules. Un observateur est synthétisé afin d’estimer les tensions des capacités. Les deux procédures de diagnostic sont appliquées pour la détection d’une variation des valeurs des capacités et le diagnostic de cellules bloquées. Enfin, une commande binaire pour le convertisseur est proposée. L’application de cette stratégie permettra, par la suite, la commande tolérante aux fautes du convertisseur. / This thesis deals with the diagnosis and the observation of a large class of switched linear systems with an application to the multicellular converter. The objective is to provide solutions for non-observable subsystems in the classical sense which can be influenced by continuous or discrete faults. After presenting a state of the art for the observation and diagnostic techniques for switched systems, the report is divided into two parts. The first part provides, in one hand, a strategy for the discrete and continuous states estimation for linear switched system with unknown input. A hybridobserver based on higher order sliding mode is developed. On the other hand, two diagnostic procedures are presented. The first one combines a hybrid observer and a discrete diagnoser to detect a continuous fault. In the second one, an active diagnosis is defined based on the testing theory to detect and isolate a discrete fault. In the second part of this thesis, the different steps to create a multicellular converter are detailed. Then, a chapter is dedicated to the validation of the theoretical approaches for the observation and diagnosis of the three cells converter. An observer is designed to estimate the capacitor voltages. The two proposed diagnosis procedures are applied to detect a change in the capacitance values and to diagnose locked cells. Finally, a binary control for the converter is proposed. The implementation of this strategy will allow, in the future, the fault tolerant control of the converter.

L'objectif de cette thèse est de proposer l'approche la plus appropriée afin de minimiser le coût d'intégration de petite éolienne dans un micro-réseau DC urbain. Une petit éolienne basé sur un machine synchrone à aimant permanent (MSAP) est considéré à étudier. Un état de l'art concernant les énergies renouvelables, micro-réseau DC, et la production d'énergie éolienne, est fait. Comme le capteur mécanique de cette structure est relativement d'un coût élevé, les différents types de contrôle pour un système de conversion éolienne sont présentés afin de choisir une structure active de conversion d'énergie et un MSAP sans capteur. Par conséquent, un estimateur de vitesse/position est nécessaire pour contrôler le système. Ainsi, les méthodes différentes proposées dans la littérature sont considérées et classifiées à étudier dans les détails, puis les plus efficaces et largement utilisés sont à vérifier dans la simulation et expérimentalement pour le système étudié. Les méthodes choisies sont: estimation de la flux de rotor avec boucle à verrouillage de phase (PLL), observateur à mode glissement (SMO), observateur de Luenberger d'ordre réduit, et filtre de Kalman étendu (EKF). Face à d'autres méthodes, l'estimateur basé sur un modèle EKF permet une commande sans capteur dans une large plage de vitesse et estime la vitesse de rotation avec une réponse rapide. Le réglage des paramètres EKF est le problème principal à sa mise en œuvre. Par conséquent, pour résoudre ce problème, la thèse présente une méthode adaptative, à savoir réglage-adaptatif d’EKF. En conséquence, et grâce à cette approche, le coût total du système de conversion est réduite et la performance est garantie et optimisée. / The aim of this thesis is to propose the most appropriate approach in order to minimize the cost of integration of a wind generator into a DC urban microgrid. A small-scale wind generator based on a permanent magnet synchronous machine (PMSM) is considered to be studied. A state of the art concerning the renewable energies, DC microgrid, and wind power generation is done. As the mechanical sensor for this structure is relatively of high cost, various types of wind conversion system control are presented in order to choose an energy conversion active structure and a sensorless PMSM. Therefore, a speed/position estimator is required to control the system. Thus, different methods proposed in literatures are considered and classified to be studied in details, and then the most effective and widely used ones are to be verified in simulation and experimentally for the studied system. The methods which are chosen are: rotor flux estimation with phase locked loop (PLL), sliding mode observer (SMO), Luenberger observer of reduced order, and extended Kalman filter (EKF). Facing to other methods, the EKF model-based estimator allows sensorless drive control in a wide speed range and estimates the rotation speed with a rapid response. The EKF parameters tuning is the main problem to its implementation. Hence, to solve this problem, the thesis introduces an adaptive method, i.e. adaptive-tuning EKF. As a result and grace to this approach, the total cost of conversion system is reduced and the performance is guaranteed and optimized.

[ES] Las infusiones exógenas de insulina son vitales para las personas con diabetes tipo 1 para suplir parcialmente la incapacidad del páncreas de secretar insulina. Sin embargo, las terapias intensivas actuales pueden restringir la calidad de vida de los pacientes. Los pacientes con esta enfermedad tienen que tomar decisiones constantemente sobre las dosis de insulina que lleva a la glucosa a niveles seguros. Si no lo consiguen, pueden sufrir las complicaciones crónicas y agudas derivadas de los niveles anormalmente altos o bajos de glucosa. La regulación automática de glucosa con sistemas de páncreas artificial prometía reducir la carga del autocontrol de la enfermedad al mismo tiempo que se mejoraba el tiempo en normoglucemia y se reducía la variabilidad. Sin embargo, estas promesas sólo se han cumplido parcialmente. Aunque esta tecnología mejora el control glucémico que logran las terapias tradicionales, la ingesta de alimentos y la práctica de ejercicio limitan la eficiencia de los sistemas de páncreas artificial durante el día. De hecho, los sistemas comerciales sólo pueden hacerles frente con la ayuda de los pacientes. Para compensar las ingestas, los pacientes deben anunciar el contenido de carbohidratos al sistema. Para el ejercicio, deben anunciar el inicio de la actividad o tomar medidas preventivas como modificar la referencia de glucosa o reducir la basal con mucha antelación. Estas exigencias no sólo no ayudan a reducir la carga del paciente, sino que pueden comprometer la eficiencia del sistema cuando el paciente se equivoca al estimar los carbohidratos, omite el anuncio de la comida o no puede planificar el ejercicio. Así pues, esta tesis propone nuevos métodos para eliminar el anuncio de ingestas y ejercicio lo que ayudaría a reducir la intervención del paciente en los sistemas de páncreas artificial, y, en consecuencia, mejorar la calidad de vida. Desde un punto de vista de control, las ingestas y el ejercicio pueden considerarse perturbaciones. Esta tesis explota métodos de la literatura de rechazo de perturbaciones y acomodación de fallos para lograr el objetivo de la tesis. En concreto, hay que destacar tres aplicaciones: 1) se ha desarrollado un observador super-twisting para detectar comidas no anunciadas como primer paso para su compensación; 2) se ha diseñado un observador de modos deslizantes de primer orden para la estimación de la ratio de aparición de glucosa, la cual ha sido integrada en un generador de bolos para compensar las comidas; 3) se ha empleado el principio de diseño por modelo interno para mitigar el efecto de las ingestas de alimentos y ejercicio, añadiendo sugerencias de carbohidratos de rescate a la insulina. Como resultado, las contribuciones de esta tesis allanan el camino para el desarrollo de sistemas de páncreas artificial sin anuncios, que liberen al paciente de la carga de la gestión de la diabetes. / [CA] Les infusions exògenes d'insulina són vitals per a les persones amb diabetis tipus 1 per a suplir parcialment l'incapacitat del pàncrees de secretar insulina. No obstant, les teràpies intensives actuals poden restringir la qualitat de vida dels pacients. Les persones amb aquesta malaltia han de prendre decisions contínuament sobre la dosi d'insulina que fa que la glucosa estiga en valors segurs. Si no ho aconsegueixen, poden sofrir les complicacions cròniques i agudes derivades dels nivells anormalment alts o baixos de glucosa. La regulació automàtica de glucosa amb sistemes de pàncrees artificials prometia reduir la càrrega d'autocontrol de la malaltia al mateix temps que es millorava el temps en normoglucèmia i es reduïa la variabilitat. No obstant, aquestes promeses s'han complit només parcialment. Encara que aquesta tecnologia millora el control de la glucosa, la ingesta d'aliments i la pràctica d'exercici limiten l'eficiència dels sistemes de pàncrees artificials durant el dia. De fet, els sistemes comercials només poden fer-les front amb la ajuda dels pacients. Per a compensar les ingestes, el pacients han d'anunciar al sistema la quantitat de carbohidrats. Per al exercici, han d'anunciar l'inici de l'activitat o prendre mesures preventives com modificar la referència de glucosa o reduir la infusió basal d'insulina. Aquestes exigències no només no ajuden a reduir la càrrega al pacient, sinó que poden comprometre l'eficiència del sistema quan el pacient confon la estimació dels carbohidrats, omet l'anunciament de la ingesta o no pot planificar l'exercici. Així doncs, aquesta tesi proposa nous mètodes per a eliminar l'anunciament d'ingestes i exercici permetent reduir així la intervenció del pacient, i, en conseqüència, millorar la qualitat de vida. Des del punt de vista de control, les ingestes i el exercici poden considerar-se pertorbacions. Aquesta tesi explota mètodes de la literatura de rebuig de pertorbacions i acomodament de fallades. En particular, es destaquen tres aplicacions: 1) s'ha desenvolupat un observador super-twisting per a detectar ingestes no anunciades com a primer pas per a la seua compensació; 2) se ha dissenyat un observador de modes lliscants per a estimar el rati d'aparició de glucosa de la ingesta, el qual s'ha integrat en un generador de bols d'insulina per compensar les ingestes; 3) se ha empleat el principi de disseny per model intern per a mitigar l'efecte de les ingestes i exercici, incorporant recomanacions de carbohidrats a la insulina. Com a resultat, les contribucions d'aquesta tesi obren el camí per a desenvolupar sistemes de pàncrees artificial sense anunciaments, que alliberen al pacient de la càrrega de la gestió de la diabetis. / [EN] Exogenous insulin infusions are vital for people with type 1 diabetes to partially make up for the inability of the pancreas to secrete insulin. However, current intensive therapies may restrict patients' quality of life. People with this disease have to constantly make decisions about insulin doses to bring glucose levels to a safe range. If unsuccessful, they may suffer from chronic and acute complications related to abnormally high and low glucose values. The automatic regulation of glucose with artificial pancreas systems promised to reduce patients' self-control burdens while improving time in normoglycemia and decreasing variability. However, these promises are fulfilled only partially. Although this technology outperforms the glycemic outcomes achieved by conventional therapies, meal intake and physical activity limit its daytime performance. Indeed, commercially available systems only can handle them with the support of the patients. For meals, patients must announce the carbohydrate content to the system. For exercise, they must notify the activity or take preventive actions like changing glucose setpoint or decreasing basal well ahead before exercise. These demands do not help reduce the patient's burden. They even can compromise the system performance when the patient misestimates the carbohydrate content, omits the meal announcement, or cannot plan the exercise event. Hence, this thesis proposes new methods to eliminate the need for meal and exercise announcements, thus reducing patient intervention in artificial pancreas systems for a better quality of life. From a control perspective, meals and exercise can be regarded as disturbances; therefore, this thesis exploits methods from the disturbance rejection and fault accommodation literature to achieve the thesis goal. Specifically, the following three applications must be highlighted: 1) a super-twisting-based residual generator has been developed to detect unannounced meals as the first step to their compensations; 2) a sliding-mode disturbance observer has been designed to estimate the glucose meal appearance, which is fed into a bolusing algorithm to compensate the meals; 3) the internal model principle is employed to mitigate the effects of meal intakes and exercise, supplementing insulin infusions with carbohydrate recommendations. As a result, the contributions of this thesis pave the way for the development of announcement-free artificial pancreas systems, releasing patients from the burden of diabetes management. / This work was partially supported by: grant DPI2016-78831-C2-1-R funded by MCIN/AEI/10.13039/501100011033 and by “ERDF A way of making Europe”; grant PID2019-107722RB-C21 funded by MCIN/AEI/10.13039/501100011033; and grant ACIF/2017/021 funded by the Generalitat Valenciana through European Social Funds (FSE). Also, the grant BEFPI/2019/077, funded by the Generalitat Valenciana through FSE funds, supported a 7-month research stay at Physiological da University (Óbudai Egyetem, Óbuda, Budapest, Hungary) / Sala Mira, I. (2023). New Contributions Towards Meal and Exercise Announcement-Free Artificial Pancreas Systems [Tesis doctoral]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/192684