This thesis is an effort to develop generalized dynamic models for systems with time-varying time delays. Unlike the simple time-delay model characterized by a transportation lag in the case of a fixed time delay, time-varying delays exhibit quite different characteristics, making the development of easy to use models a difficult endeavor. First an algorithm is developed to predict the actual input-output behavior when the input signal is directly fed into a device that characterizes the time-varying delay. Input-output behaviour generated with this algorithm serves as the truth model for subsequent approximate model development. Simulation results for different classes of delay and different inputs were obtained using the truth model. The input functions were limited to steps, ramps and sinusoids. This limited class of inputs and delays defines the scope of this thesis and the results are to be interpreted as such. The methodology adopted to identify the basic underpinnings of models was system identification where input-output data came from the truth model. Models for the aforementioned classes of inputs and delays were then derived using elementary system identification tools. These models were then carefully analyzed to extract trends by changing the delay parameter. A satisfactory trend was observed in the case of linearly varying time delay. A generalized model for the linearly varying time delay with step and polynomial inputs was developed. An attempt was also made at developing a generalized model for sinusoidally time-varying time delays. This study proposes a model for linearly time-varying time delay, whose structure is not surprisingly also dependent on the class of inputs. It is shown that the derived model reduces to the well known model in the case of a fixed delay.

This thesis is an effort to develop generalized dynamic models for systems with time-varying time delays. Unlike the simple time-delay model characterized by a transportation lag in the case of a fixed time delay, time-varying delays exhibit quite different characteristics, making the development of easy to use models a difficult endeavor. First an algorithm is developed to predict the actual input-output behavior when the input signal is directly fed into a device that characterizes the time-varying delay. Input-output behaviour generated with this algorithm serves as the truth model for subsequent approximate model development. Simulation results for different classes of delay and different inputs were obtained using the truth model. The input functions were limited to steps, ramps and sinusoids. This limited class of inputs and delays defines the scope of this thesis and the results are to be interpreted as such. The methodology adopted to identify the basic underpinnings of models was system identification where input-output data came from the truth model. Models for the aforementioned classes of inputs and delays were then derived using elementary system identification tools. These models were then carefully analyzed to extract trends by changing the delay parameter. A satisfactory trend was observed in the case of linearly varying time delay. A generalized model for the linearly varying time delay with step and polynomial inputs was developed. An attempt was also made at developing a generalized model for sinusoidally time-varying time delays. This study proposes a model for linearly time-varying time delay, whose structure is not surprisingly also dependent on the class of inputs. It is shown that the derived model reduces to the well known model in the case of a fixed delay.

In this dissertation, we will investigate the global stability for some uncertain neural networks with multiple time-varying delays. These well-known neural networks include delayed cellular neural networks (DCNNs), delayed bidirectional associative memory neural networks (DBAMNNs), and delayed Cohen-Grossberg neural networks (DCGNNs). Delay-dependent and delay-independent criteria will be proposed to guarantee the robust stability of these uncertain delayed neural networks via linear matrix inequality (LMI) approach. Three types of uncertainties on feedback and delayed feedback matrices in these uncertain delayed neural networks will be considered in this study, namely uncertainties with structured perturbation, norm-bounded unstructured perturbation, and interval perturbation. Some numerical examples will be given to illustrate the effectiveness of our results. Some comparisions are made to show that our results are better than some results in recent literature.

The thesis focuses on the design of a control strategy for safety-critical remote teleoperation. The main goal is to make the controlled system track the desired velocity specified by a human operator while avoiding obstacles despite communication delays. Different methods adopting Control Barrier Functions (CBFs) and Model Predictive Control (MPC) have been explored and tested. In this combination, CBFs are used to define the safety constraints the system has to respect to avoid obstacles, while MPC provides the framework for filtering the desired input by solving an optimization problem. The resulting input is sent to the remote system, where appropriate low-level velocity controllers translate it into system-specific commands. The main novelty of the thesis is a method to make the CBFs robust against the uncertainties affecting the system’s state due to network delays. Other techniques are investigated to improve the quality of the system information starting from the delayed one and to formulate the optimization problem without knowing the specific dynamic of the controlled system. The results show how the proposed method successfully solves the safetycritical teleoperation problem, making the controlled systems avoid obstacles with different types of network delay. The controller has also been tested in simulation and on a real manipulator, demonstrating its general applicability when reliable low-level velocity controllers are available. / Avhandlingen fokuserar på utformningen av en kontrollstrategi för säkerhetskritisk fjärrstyrd teleoperation. Huvudmålet är att få det kontrollerade systemet att följa den önskade hastigheten som specificeras av en mänsklig operatör samtidigt som hinder undviks trots kommunikationsfördröjningar. Olika metoder som använder Control Barrier Functions (CBFs) och Model Predictive Control har undersökts och testats. I denna kombination används CBFs för att definiera de säkerhetsbegränsningar som systemet måste respektera för att undvika hinder, medan MPC utgör ramverket för filtrering av den önskade indata genom att lösa ett optimeringsproblem. Den resulterande indata skickas till fjärrsystemet, där lämpliga hastighetsregulatorer på låg nivå översätter den till systemspecifika kommandon. Den viktigaste nyheten i avhandlingen är en metod för att göra CBFs robust mot de osäkerheter som påverkar systemets tillstånd på grund av nätverksfördröjningar. Andra tekniker undersöks för att förbättra kvaliteten på systeminformationen med utgångspunkt från den fördröjda informationen och för att formulera optimeringsproblemet utan att känna till det kontrollerade systemets specifika dynamik. Resultaten visar hur den föreslagna metoden framgångsrikt löser det säkerhetskritiska teleoperationsproblemet, vilket gör att de kontrollerade systemen undviker hinder med olika typer av nätverksfördröjningar. Styrningen har också testats i simulering och på en verklig manipulator, vilket visar dess allmänna tillämpbarhet när tillförlitliga lågnivåhastighetsregulatorer finns tillgängliga.

Ce travail propose de nouvelles structures de contrôle pour la téléopération bilatérale à travers des réseaux de communication non dédiés. L’enjeu est donc de concevoir et calculer des structures de commande garantissant la stabilisation et un bon degré de performance en termes de synchronisation (suivi des positions et vitesse) et de transparence (ressenti des forces) sous les retards variables et asymétriques.Nous faisons tout d’abord un tour d’horizon des recherches récentes dans le domaine des systèmes de téléopération et de leurs caractéristiques. Puis, nous considérons des modèles linéaires à plusieurs retards variables pour lesquels nous proposons une approche d’analyse de stabilité par fonctionnelles de Lyapunov-Krasovskii et contrôle robuste H [infinity symbol . Ensuite, trois structures de téléopération seront proposées en temps continu, la comparaison de ces architectures montre que, pour un retard de réseau maximum donné ou calculé, toutes garantissent un suivi de position et vitesse. Les deux dernières, qui utilisent les forces mesurées ou estimées de l’opérateur humain et de l’environnement, garantissent de plus un suivi en force. Au final, la troisième structure (avec proxy) présente la meilleure performance, même si elle demande un peu plus de calcul. Puis, afin d’analyser et d’améliorer les performances de la troisième structure pour des modèles encore plus réalistes, une étude est menée en temps discret, mais aussi sur un modèle non linéaire ou non stationnaire sous perturbations bornées en norme. L’implantation sur la plate-forme est décrite dans un quatrième et dernier chapitre, et puis l’analyse des résultats expérimentaux est alors menée / This PhD thesis is dedicated to the control scheme design of the bilateral teleoperation under asymmetric communication channels: the stabilization and a high-level performance under asymmetric time-varying delays and the perturbations of the human operator and environment. After a review of the recent researches and their features in the field of teleoperation system, a less conservative Lyapunov-Krasovskii functional together with H [infinity symbol] control theory has been applied to linear time delay systems, and then the LMI theorems have been obtained in order to calculate the controllers in the control schemes.Firstly, three novel teleoperation control schemes have been presented. Comparing three architectures, all of them guaranteed the stability and the position tracking thanks to the position/velocity information. Force-reflecting control scheme without or with proxy, in addition, ensured the force tracking by using the estimated/measured force of the human operator and the environment. Here, the control scheme with the proxy got a better performance. Secondly, a discrete-time approach has been developed to analyze the force-reflecting control scheme with proxy and obtain a better system performance. Besides, more general systems with time-varying uncertainties (the polytopic-type uncertainties and the norm-bounded model uncertainties) have been considered. Finally, the experimental test-bench and the real system implementation have been designed, which involved the identification and linearizing control of the subsystems (the master/slave robots). The experimental results have illustrated the effectiveness of the approaches proposed in this thesis

En este trabajo se aborda el problema de controlar un proceso cuya salida se muestrea de forma irregular. Para ello se propone utilizar un predictor que estima las salidas del proceso en instantes regulares de tiempo más un controlador convencional que calcula la acción de control a partir de las estimaciones del predictor (técnica conocida como control inferencial). La predicción consiste en estimar las variables de salida que se desean controlar a partir de las mediciones realizadas con diversos sensores utilizando para ello un modelo matemático del proceso. El filtro de Kalman permite hacer la predicción de forma óptima si las perturbaciones tienen una distribución gaussiana de media cero, pero con el inconveniente de requerir un elevado coste computacional cuando se utilizan diferentes sensores con retardos temporales variantes. En este trabajo se propone una estrategia de predicción alternativa de bajo coste computacional cuyo diseño se basa en el conocimiento de la disponibilidad de mediciones y de los retardos (del proceso, del sistema de medición o del sistema de transmisión de datos) y de la naturaleza de las perturbaciones. Los predictores propuestos minimizan el error de predicción frente al muestreo aleatorio con retardos variantes, perturbaciones, ruido de medida, error de modelado, retardos en la acción de control e incertidumbre en los tiempos de medición. Las diferentes estrategias de diseño que se proponen se clasifican según el tipo de información que se dispone de las perturbaciones y del coste computacional requerido. Se han planteado los diseños para sistemas monovariables, multivariables, lineales y no lineales. Asimismo, también se ha elaborado una forma más eficiente de incluir mediciones escasas con retardo en el filtro de Kalman, con el objetivo de reducir el coste computacional de la predicción. En este trabajo se demuestra que los sistemas de control inferencial que utilizan los predictores propuestos cumplen con el principio de sep / Peñarrocha Alós, I. (2006). Sensores virtuales para procesos con medidas escasas y retardos temporales [Tesis doctoral no publicada]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/3882 / Palancia