Traditionally, the task of determining aircraft position and attitude for automatic control has been handled by the combination of an inertial measurement unit (IMU) with a Global Positioning System (GPS) receiver. In this configuration, accelerations and angular rates from the IMU can be integrated forward in time, and position updates from the GPS can be used to bound the errors that result from this integration. However, reliance on the reception of GPS signals places artificial constraints on aircraft such as small unmanned aerial vehicles (UAVs) that are otherwise physically capable of operation in indoor, cluttered, or adversarial environments. Therefore, this work investigates methods for incorporating a monocular vision sensor into a standard avionics suite. Vision sensors possess the potential to extract information about the surrounding environment and determine the locations of features or points of interest. Having mapped out landmarks in an unknown environment, subsequent observations by the vision sensor can in turn be used to resolve aircraft position and orientation while continuing to map out new features. An extended Kalman filter framework for performing the tasks of vision-based mapping and navigation is presented. Feature points are detected in each image using a Harris corner detector, and these feature measurements are corresponded from frame to frame using a statistical Z-test. When GPS is available, sequential observations of a single landmark point allow the point's location in inertial space to be estimated. When GPS is not available, landmarks that have been sufficiently triangulated can be used for estimating vehicle position and attitude. Simulation and real-time flight test results for vision-based mapping and navigation are presented to demonstrate feasibility in real-time applications. These methods are then integrated into a practical framework for flight in GPS-denied environments and verified through the autonomous flight of a UAV during a loss-of-GPS scenario. The methodology is also extended to the application of vehicles equipped with stereo vision systems. This framework enables aircraft capable of hovering in place to maintain a bounded pose estimate indefinitely without drift during a GPS outage.

A new suboptimal guidance law design approach for aerospace vehicles is proposed in this thesis, followed by an advanced control design for supersonic air-breathing engines. The guidance law is designed using the newly developed Generalized Model Predictive Static Programming (G-MPSP), which is based on the continuous time nonlinear optimal control framework. The key feature of this technique is one-time backward propagation of a small-dimensional weighting matrix dynamics, which is used to update the entire control history. This key feature, as well as the fact that it leads to a static optimization problem, lead to its computational efficiency. It has also been shown that the existing model predictive static programming (MPSP), which is based on the discrete time framework, is a special case of G-MPSP. The G-MPSP technique is further extended to incorporate ‘input inequality constraints’ in a limited sense using the penalty function philosophy. Next, this technique has been developed also further in a ‘flexible final time’ framework to converge rapidly to meet very stringent final conditions with limited number of iterations. Using the G-MPSP technique in a flexible final time and input inequality constrained formulation, a suboptimal guidance law for a solid motor propelled carrier launch vehicle is successfully designed for a hypersonic mission. This guidance law assures very stringent final conditions at the injection point at the end of the guidance phase for successful beginning of the hypersonic vehicle operation. It also ensures that the angle of attack and structural load bounds are not violated throughout the trajectory. A second-order autopilot has been incorporated in the simulation studies to mimic the effect of the inner-loops on the guidance performance. Simulation studies with perturbations in the thrust-time behaviour, drag coefficient and mass demonstrate that the proposed guidance can meet the stringent requirements of the hypersonic mission. The G-MPSP technique in a fixed final time and input inequality constrained formulation has also been used for optimal guidance of an aerospace vehicle propelled by supersonic air-breathing engine, where the resulting thrust can be manipulated by managing the fuel flow and nozzle area (which is not possible in solid motors). However, operation of supersonic air-breathing engines is quite complex as the thrust produced by the engine is a result of very complex nonlinear combustion dynamics inside the engine. Hence, to generate the desired thrust, accounting for a fairly detailed engine model, a dynamic inversion based nonlinear state feedback control design has been carried out. The objective of this controller is to ensure that the engine dynamically produces the thrust that tracks the commanded value of thrust generated from the guidance loop as closely as possible by regulating the fuel flow rate. Simultaneously, by manipulating throat area of the nozzle, it also manages the shock wave location in the intake for maximum pressure recovery with sufficient margin for robustness. To filter out the sensor and process noises and to estimate the states for making the control design operate based on output feedback, an extended Kalman filter (EKF) based state estimation design has also been carried out and the controller has been made to operate based on estimated states. Moreover, independent control designs have also been carried out for the actuators so that their response can be faster. In addition, this control design becomes more challenging to satisfy the imposed practical constraints like fuel-air ratio and peak combustion temperature limits. Simulation results clearly indicate that the proposed design is quite successful in assuring the desired performance of the air-breathing engine throughout the flight trajectory, i.e., both during the climb and cruise phases, while assuring adequate pressure margin for shock wave management.

Cette thèse est consacrée à la restauration et l'estimation des paramètres par filtrage dans les modèles de chaîne de Markov cachée classique, couple et triplet à sauts Markoviens. Nous proposons deux nouvelles méthodes d'approximation dans le cas des systèmes linéaires gaussiens à sauts Markoviens. La première est fondée sur l'utilisation des chaînes de Markov cachées par du bruit à mémoire longue, on obtient alors une méthode " partiellement non supervisée" dans la quelle certains paramètres, peuvent être estimés en utilisant une version adaptative de l'algorithme EM ou ICE, les résultats obtenus sont encourageant et comparables avec les méthodes classiquement utilisées du type (Kalman/Particulaire). La deuxième exploite l'idée de ne garder à chaque instant que les trajectoires les plus probables; là aussi, on obtient une méthode très rapide donnant des résultats très intéressants. Nous proposons par la suite deux familles de modèles à sauts qui sont originaux. la première est très générale où le processus couple composé du processus d'intérêt et celui des observations conditionnellement aux sauts, est une chaîne de Markov cachée, et nous proposons une extension du filtrage particulaire à cette famille. La deuxième, est une sous famille de la première où le couple composé de la chaîne des sauts et le processus d'observations est Markovien dans ce dernier cas le filtrage optimal exact est possible avec une complexité linéaire dans le temps. L'utilisation de la deuxième famille en tant qu'approximation de la première est alors étudiée et les résultats exposés dans ce mémoire semblent très encourageants / This thesis is devoted to the restoration problem and the parameter estimation by filtering in the traditional hidden Markov chain model, couple and triplet with Markovian jumps. We propose two new approximate methods in the case of Gaussian linear systems with Markovian jumps. first is founded to use the hidden Markov chains by noise with long memory, we obtains a method " partially not supervised" some parameters, can be estimated by using an adaptive version of EM or ICE algorithm, the results obtained are encouraging and comparable with the methods used classically (Kalman/Particle). The second one exploits idea to keep at every moment only the most probable trajectories; we obtains a very fast method giving very interesting results. Then we propose two families of models to jumps which are original. The first one is very general where the process couples made up of the hidden and the observations process conditionally to the jumps, are a hidden Markov chain, and we propose an extension of particulate filtering to this family. The second is under family of the first, where the couple made up of the jumps and the observations process is Markovian, in this last case exact optimal filtering is possible with a linear complexity in time. Using of the second family to approach the first one is studied and the results exposed in this memory seem very encouraging

This thesis presents an investigation of the possibility of using the fixed-point arithmetic in the inertial navigation systems, which use the local level navigation frame mechanization equations. Two square root filtering methods, the Potter's square root Kalman filter and UD factorized Kalman filter, are compared with respect to the conventional Kalman filter and its Joseph's stabilized form. The effect of rounding errors to the Kalman filter optimality and the covariance matrix or its factors conditioning is evaluated for a various lengths of the fractional part of the fixed-point computational word. Main contribution of this research lies in an evaluation of the minimal fixed-point arithmetic word length for the Phi-angle error model with noise statistics which correspond to the tactical grade inertial measurements units.

No description available.

<p>An emerging technology in the field of transient thermal management is thermal energy storage, or TES, which enables temporary, on-demand heat rejection via storage as latent heat in a phase-change material.  Latent TES devices have enabled advances in many thermal management applications, including peak load shifting for reducing energy demand and cost of HVAC systems and providing supplemental heat rejection in transient thermal management systems.  However, the design of a transient thermal management system with integrated storage comprises many challenges which are yet to be solved.  For example, design approaches and performance metrics for determining the optimal dimensions of the TES device have only recently been studied.  Another area of active research is estimation of the internal temperature state of the device, which can be difficult to directly measure given the transient nature of the thermal storage process.  Furthermore, in contrast to the three main functions of a thermal-fluid system--heat addition, thermal transport, and heat rejection--thermal storage introduces the need for active, real-time control and automated decision making for managing the operation of the thermal storage device. </p> <p>In this thesis, I present the design process for integrating thermal energy storage into a single-phase thermal management system for rejecting transient heat loads, including design of the TES device, state estimation and control algorithm design, and validation in both simulation and experimental environments. Leveraging a reduced-order finite volume simulation model of a plate-fin TES device, I develop a design approach which involves a transient simulation-based design optimization to determine the required geometric dimensions of the device to meet transient performance objectives while maximizing power density.  The optimized TES device is integrated into a single-phase thermal-fluid testbed for experimental testing.  Using the finite volume model and feedback from thermocouples embedded in the device, I design and experimentally validate a state estimator based on the state-dependent Riccati equation approach for determining the internal temperature distribution to a high degree of accuracy.  Real-time knowledge of the internal temperature state is critical for making control decisions; to manage the operation of the TES device in the context of a transient thermal management system, I design and test, both in simulation and experimentally, a logic-based control strategy that uses fluid temperature measurements and estimates of the TES state to make real-time control decisions to meet critical thermal management objectives. Together, these advances demonstrate the potential of thermal energy storage technology as a component of thermal management systems and the feasibility of logic-based control strategies for real-time control of thermal management objectives.</p>

<p>    </p> <p>This thesis studies networked Gaussian linear control systems with random delays. Networked control systems is a popular topic these years because of their versatile applications in daily life, such as smart grid and unmanned vehicles. With the development of these systems, researchers have explored this area in two directions. The first one is to derive the inherent rate-cost relationship in the systems, that is the minimal transmission rate needed to achieve an arbitrarily given stability requirement. The other one is to design achievability schemes, which aim at using as less as transmission rate to achieve an arbitrarily given stability requirement. In this thesis, we explore both directions. We assume the sensor-to-controller channels experience independently and identically distributed random delays of bounded support. Our work separates into two parts. In the first part, we consider networked systems with only one sensor. We focus on deriving a lower bound, R_{LB}(D), of the rate-cost tradeoff with the cost function to be E{| <strong>x^</strong>T<strong>x </strong>|} ≤ D, where <strong>x </strong>refers to the state to be controlled. We also propose an achievability scheme as an upper bound, R_{UB}(D), of the optimal rate-cost tradeoff. The scheme uses lattice quantization, entropy encoder, and certainty-equivalence controller. It achieves a good performance that roughly requires 2 bits per time slot more than R_{LB}(D) to achieve the same stability level. We also generalize the cost function to be of both the state and the control actions. For the joint state-and-control cost, we propose the minimal cost a system can achieve. The second part focuses on to the covariance-based fusion scheme design for systems with multiple > 1 sensors. We notice that in the multi-sensor scenario, the outdated arrivals at the controller, which many existing fusion schemes often discard, carry additional information. Therefore, we design an implementable fusion scheme (CQE) which is the MMSE estimator using both the freshest and outdated information at the controller. Our experiment demonstrates that CQE out-performances the MMSE estimator using the freshest information (LQE) exclusively by achieving a 15% smaller average L2 norm using the same transmission rate. As a benchmark, we also derive the minimal achievable L2 norm, Dmin, for the multi-sensor systems. The simulation shows that CQE approaches Dmin significantly better than LQE. </p>

[ES] Esta tesis de doctorado está dedicada al problema de descomposición de señales no estacionarias en componentes modales, entendida como componentes oscilatorias independientes, con amplitud y fase dependientes del tiempo. Para este fin, se propone un enfoque metodológico basado en representaciones en espacio de estados diagonales en bloques. Una contribución teórica primaria de esta tesis consiste en demostrar que la respuesta de un sistema de espacio de estados diagonal en bloques puede ser representada en una forma modal con amplitudes y frecuencias dependientes del tiempo. Subsecuentemente, construyendo sobre este resultado, un marco de trabajo basado en filtros de Kalman se propone para la descomposición modal de señales no estacionarias. Como resultado, una familia de métodos paramétricos para la descomposición modal de señales no estacionarias univariadas y multivariadas basadas en representaciones de espacio de estados diagonales en bloques y filtros de Kalman ha sido postulada. La representación básica está construida en bloques de segundo orden, cada uno de los cuales representa los componentes en fase y en cuadratura de un único componente oscilatorio no estacionario. Así, la respuesta total es construida como la suma ponderada de cada uno de estos modos. La identificación de estos modelos requiere la estimación conjunta de las trayectorias y los parámetros modales dependientes del tiempo, así como los hiperparámetros del modelo, constituidos por la matriz de mezcla de modos, las matrices de covarianza del vector de estados, de parámetros y del ruido de medición, y las condiciones iniciales. Para este propósito, un algoritmo de Expectación-Maximización ha sido adaptado como parte de esta tesis. La metodología obtenida es entonces evaluada en la descomposición y eliminación de ruido de registros electrocardiográficos (ECG), los cuales consisten en componentes no-estacionarias pseudo-periódicas y son susceptibles a diferentes tipos de interferencias. La estructura de estas señales las hace susceptibles a las descomposiciones modales basadas propuestas en esta tesis. A diferencia de otros métodos populares de descomposición de señales, las descomposiciones obtenidas con la metodología propuesta proveen componentes oscilatorios con interpretabilidad física y que proveen resultados consistentes para señales multivariadas, como en el caso de registros de ECG con múltiples derivaciones. Otra estrategia que se desarrolló en este proyecto investigativo lo constituye la aplicación de la transformada delta u operador de Euler al filtro de Kalman, esto condujo a resultados de alta precisión en la extracción de componentes de banda angosta. La metodología propuesta constituye una herramienta confiable para la descomposición modal en línea de señales no estacionarias multicomponentes, con resultados excelentes / [CA] Esta tesi de doctorat està dedicada al problema de descomposició de senyals no-estacionaris en components modals, entesa com a components oscil·latòries independents amb amplitud i fase dependents del temps. Per a este fi, es proposa un enfocament metodològic basat en representacions en espai d'estats diagonals en blocs. Una contribució teòrica primària d'esta tesi consistix a demostrar que la resposta d'un sistema d'espai d'estats diagonal en blocs pot ser representada en una forma modal amb amplituds i freqüències dependents del temps. Subseqüentment, construint sobre este resultat, un marc de treball basat en filtres de Kalman es proposa per a la descomposició modal de senyals no estacionaris. Com a resultat, una família de mètodes paramètrics per a la descomposició modal de senyals no estacionaris univariadas i multivariades basades en representacions d'espai d'estats diagonals en blocs i filtres de Kalman ha sigut postulada. La representació bàsica està construïda en blocs de segon ordre, cadascun dels quals representa els components en fase i en quadratura d'un únic component oscil·latori no estacionari. Així, la resposta total és construïda com la suma ponderada de cadascun d'estos modes. La identificació d'estos models requerix l'estimació conjunta de les trajectòries i els paràmetres modals dependents del temps, així com els hiperparámetros del model, constituïts per la matriu de mescla de modes, les matrius de covariància del vector d'estats, de paràmetres i del soroll de mesurament, i les condicions inicials. Per a este propòsit, un algorisme d'Expectació-Maximització ha sigut adaptat com a part d'esta tesi. La metodologia obtinguda és llavors avaluada en la descomposició i eliminació de soroll de registres electrocardiogràfics (ECG), els quals consistixen en components no-estacionàries pseudo-periòdiques i són susceptibles a diferents tipus d'interferències. L'estructura d'estos senyals les fa susceptibles a les descomposicions modals basades propostes en esta tesi. A diferència d'altres mètodes populars de descomposició de senyals, les descomposicions obtingudes amb la metodologia proposada proveïxen components oscil·latoris amb interpretabilidad física i que proveïxen resultats consistents per a senyals multivariats, com en el cas de registres d'ECG amb múltiples derivacions. Una altra estratègia que es va desenvolupar en este projecte investigativo el constituïx l'aplicació de la transformada delta o operador d'Euler al filtre de Kalman, això va conduir a resultats d'alta precisió en l'extracció de components de banda estreta. La metodologia proposada constituïx una eina de confiança per a la descomposició modal en línia de senyals no estacionaris multicomponents, amb resultats excel·lents. / [EN] This PhD thesis is devoted to the problem of the decomposition of non-stationary signals in modal components, understood as independent oscillatory components with time-dependent amplitude and frequency. To this end, a methodological approach based on diagonal time-dependent state space models is postulated. A primary theoretical contribution of this work is to demonstrate that the response of a system in diagonal time-dependent state space form can be cast in a modal form characterized by time-dependent amplitudes and frequencies. Subsequently, building up on this result, a Kalman filter based framework for non-stationary modal decomposition is proposed. As a result, a family of parametric modal decomposition methods is postulated for univariate and multivariate non-stationary signals based on block-diagonal time-dependent state space representations and Kalman filtering/smoothing. The representation is built upon second order blocks, each representing the in-phase and quadrature components of a single non-stationary oscillatory component. The total response is then constructed as the weighted sum of each of these modes. Accordingly, the model identification involves the joint estimation of the modal trajectories and the time-dependent modal parameters, along with the model hyperparameters, constituted by the mode mixing matrix, the state, parameter and noise covariances, and initial conditions. A tailored Expectation-Maximization algorithm is designed for this purpose as part of this thesis. The obtained methodology is assessed in the decomposition and denoising of electrocardiographic (ECG) signals, which consist of pseudo-periodic non-stationary signals and are susceptible to significant interference. The ECG signal structure makes them amenable to the proposed non-stationary modal decompositions. In contrast to other popular non-stationary signal decomposition methods, the proposed method provides a physically meaningful decomposition of oscillatory components, with consistent results for multivariate signals, such as multi-lead ECG records. Another strategy that was developed in this research project is the application of the delta transform or Euler operator to the Kalman filter, which led to highly precise results in extracting narrowband components. The proposed methodology constitutes a reliable tool for on-line modal decomposition of multi-component non-stationary signals, with results comparable and even better than other state-of-the-art methods. / Avendaño, LE. (2024). Método de descomposición modal no estacionaria basado en representación de espacio de estados con aplicación al análisis de señales ECG [Tesis doctoral]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/211185

Indiana University-Purdue University Indianapolis (IUPUI) / Lithium ion (Li-ion) batteries have become integral parts of our lives; they are widely used in applications like handheld consumer products, automotive systems, and power tools among others. To extract maximum output from a Li-ion battery under optimal conditions it is imperative to have access to the state of the battery under every operating condition. Faults occurring in the battery when left unchecked can lead to irreversible, and under extreme conditions, catastrophic damage. In this thesis, an adaptive fault diagnosis technique is developed for Li-ion batteries. For the purpose of fault diagnosis the battery is modeled by using lumped electrical elements under the equivalent circuit paradigm. The model takes into account much of the electro-chemical phenomenon while keeping the computational effort at the minimum. The diagnosis process consists of multiple models representing the various conditions of the battery. A bank of observers is used to estimate the output of each model; the estimated output is compared with the measurement for generating residual signals. These residuals are then used in the multiple model adaptive estimation (MMAE) technique for generating probabilities and for detecting the signature faults. The effectiveness of the fault detection and identification process is also dependent on the model uncertainties caused by the battery modeling process. The diagnosis performance is compared for both the linear and nonlinear battery models. The non-linear battery model better captures the actual system dynamics and results in considerable improvement and hence robust battery fault diagnosis in real time. Furthermore, it is shown that the non-linear battery model enables precise battery condition monitoring in different degrees of over-discharge.