Search

National audience; Cet article présente un observateur fonctionnel pour des systèmes linéaires et un estimayeur de fautes en convergence en temps fini. La stabilité de l'observateur est presentée et sa performance est demontréeà partir de similations numeriques. L'originalité de cet observateur fonctionnel est sa capabilité de convergir en un temps donné et garantir la correcte estimation de la faute en une periode précise de temps.

International audience; In this work, a Model-based Prognosis algorithm to predict the Flight Endurance (FE) and the Remaining Mission Time (RMT) for a class of multirotor UAVs powered by Lithium Polymer (Li-Po) batteries is presented. A Safety Voltage Threshold (SVT) in the State of Charge (SoC) of the battery is established before reaching of End of Discharge (EoD), and the SVT is defined taking into account the existing relationship between the voltage and the SoC of the battery. The FE is predicted by forecasting when the SVT is reached and the RMT is computed as the difference between the FE and the actual time. The prediction is developed in three sequential steps: 1) estimation of the battery SoC, 2) propagation and prediction of the estimated SoC in the future to reach the SVT, and 3) updating the prediction. The Model-based Prognosis algorithm is based on the mathematical model of the UAV propulsion system, which is made up by a set of BrushLess DC motors and a Li-Po battery. The effectiveness of the proposed algorithm was tested in simulation and the results obtained demonstrate the efficacy of the proposed method to accurately predict the FE and RMT during the development of a mission.

National audience; La mesure ou l’estimation exacte de l’état de charge (State-of-Charge - SoC) d’une batterie Li-Po qui alimente les véhicules volant sans pilote (Unmanned Aerial Vehicle - UAV) durant leurs vols permet de connaitre les limites de leurs missions. Ce papier présente une méthodologie d’estimation de l’état de charge d’une batterie Li-Po à partir d’un filtre de Kalman étendu (EKF) durant le vol d’un UAV. Nous considérons un UAV de type hexarotor (ou hexacoptère) pour lequel la puissance électrique consommée sera calculée et l’état de charge de la batterie sera estimée. Finalement, une stratégie basée sur l’estimation de l’état de charge de la batterie est présentée de manière à prédire la fin de la décharge (End-of-Discharge – EoD) durant la mission. Des résultats de simulation présentent l’efficacité de la méthode proposée.

International audience; The present work proposes a new approach for observer design for a general class of state affine nonlinear systems with discrete-time measurements. A new high gain observer design is developed and analyzed under insightful conditions. This result is achieved by considering a persistent excitation condition that can be validated on-line. The algorithm is applied to an induction motor and validated with simulations.

International audience; The present paper considers the H-infinity generalized dynamic observer-based control for uncertain descriptor systems by using a generalized dynamic observer (GDO), which is more general than the proportional observer (PO) and the proportional-integral observer (PIO). The objective is to stabilize an uncertain descriptor system that normally is unstable by using the estimates given by the GDO while a given level of disturbance attenuation is assured. The stability conditions for the existence of this H-infinity GDO-based control are given in terms of a set of linear matrix inequalities (LMIs). A numerical example is given to show the present approach.

National audience; The present paper considers the H-infinity observers design for linear parameter varying (LPV) singular systems by introducing a new observer structure named generalized dynamic observer (GDO), which is more general that the proportional observer (PO) and the proportional-integral observer (PIO). The objective is to design a GDO such that the resulting error dynamics is stable and a given level of disturbance attenuation is assured. The stability conditions for the existence of this H-infinity GDO are given in terms of a set of linear matrix inequalities (LMIs). A numerical example is given to show the present approach.

International audience; This work is dedicated to the design of a robust fault detection system for an anaerobic bioreactor modeled as a Descriptor Linear Parameter Varying system (D-LPV) system with unmeasurable gain scheduling functionss. The main goals are: first, an accurate D-LPV representation of the nonlinear model is obtained by considering the sector nonlinearity modeling approach. Second, a LPV observer is designed to estimate the bioprocess states variables. The observer considers the gains scheduling functions depending on estimated states due to the lack of a biomass mass sensor. Third, in oder to perform fault detection and isolation, by mean of residual generation, a bank of observers is designed. For each observer, sufficient conditions to guarantee asymptotic stability and robustness against disturbance are given by a set of feasible Linear Matrix Inequalities (LMIs). Finally, some simulations in fault-free and faulty operations are considered on the bioreactor system.

En este trabajo se presenta el diseño e implementación de un sistema de detección y localización de fallas para un proceso de copolimerización en emulsión. El enfoque propuesto para el diseño de los generadores de residuos está basado en técnicas de desacoplo usando geometría diferencial y el uso de observadores no lineales de alta ganancia, en el cual cada uno de los residuos se encuentra dedicado a la detección y a la localización de una falla en particular. Las fallas consideradas en este trabajo son de tipo aditivo, presentes en los actuadores del proceso, que representan un mal funcionamiento de las bombas de alimentación de flujo de monómeros hacia el interior del reactor. Una de las principales ventajas del método propuesto son los observadores utilizados como generadores de residuos, los cuales son fáciles de implementar y de sencilla calibración, lo que favorece su aplicación en línea. La eficacia del método se valida en simulación para el caso de un proceso de copolimerización en emulsión entre el acrilato de butilo y el estireno. In this work the design and implementation of a fault detection and isolation system for a process of copolymerization in emulsion are presented. The proposed approach used to design the residual generators is based on decoupling techniques using differential geometry and high gain nonlinear observers; every residue is dedicated to the detection and the location of a particular fault. The faults considered in this work are additive, presented in the actuators of the process and they represent a malfunction in the feeding flow pumps of the monomer towards the interior of the reactor. The observers used to generate residuals are one of the main advantages of the proposed method, these are easy to implement and have a simple calibration, which make their on-line application suitable. The effectiveness of this method was validated by simulation for the process of copolymerization in emulsion between butyl acrylate and styrene.

National audience; Dans cet article nous considérons le problème de synthèse d'un observateur pour les systèmes singuliers à temps discret en utilisant un nouveau concept d'observateurs dynamiques généralisés (ODG). L'avantage de ce nouveau concept est la structure de cet observateur, qui est plus générale que celle d'observateur proportionnel intégral (OPI) et des observateurs proportionnels (OP). Des conditions nécessaires et suffisantes de stabilité de l'observateur dynamique généralisé proposé sont données sous formes d'inégalités matricielles linéaires (LMIs).

International audience; This paper design a method for fault detection and isolation based on observers for systems modelled as Descriptor-Linear Parameter Varying (D-LPV) with Unmeasurable Scheduling Functions (USF). The first contribution of the paper is deal with the USF problem by transforming the into an uncertain D-LPV system with an estimated scheduling parameter. As a second contribution, a robust LPV observer is designed with H_{infty} performance to supply a robust state estimation against uncertainties provided by USF. Sufficient conditions to guarantee the robustness and convergence are obtained via linear matrix inequalities (LMIs). Finally, an observer bank based on robust H_{infty} observers is used to generate residuals and perform sensor fault detection and isolation. A numerical example demonstrates the effectiveness of our methods.

International audience; This paper considers the H_inf dynamical observer-based control design for descriptor systems. The goal is to design a controller such that the closed loop system is stable while the transfer function from the disturbance to the controlled output satisfies a prescribed H_inf norm level. Conditions for the solvability of this problem are obtained in terms of a set of linear matrix inequalities (LMIs). A numerical example is given to illustrate the main contribution of this paper.

International audience; This paper is concerned with the design of H observers for linear time-invariant descriptor systems. The goal is to design a dynamical observer such that the resulting dynamic estimation error is stable while the transfer function from the disturbance to this error satisfies a prescribed H norm level. Sufficient conditions for the existence and stability of these observer are given. A numerical example is given to illustrate our approach.

International audience; The paper discusses the development of a MATLAB/SCILAB toolbox for systems modeled in the LTI or LPV descriptor framework. The properties of regularity, solvability, controllability and observability are presented. Full and reduced order, proportional and proportional-integral observers are included. Some of these observers consider unknown inputs. The main contribution is provide a toolbox than can be used as auxiliary in state-estimation and fault detection based observers. These observers have been considered from few papers published recently. The fault detection and isolation can be achieved by the construction of bank of observers. These banks of observers can be built by the selection of the input/output matrices or automatically by using the proposed algorithms.

The objective of this work is to describe step-by-step how to implement an observer-based soft sensor in order to estimate process variables for which a hardware sensor is not available. The design and implementation procedure is illustrated by applying it to a counter-flow double-pipe heat exchanger. The approach used to design the nonlinear observer is based on a simplified mathematical model of the process. Numerical simulations and experiments were performed in a bench-scale pilot plant in order to validate the proposed scheme. El objetivo de este trabajo es describir paso a paso la implementación de un sensor basado en observador para estimar las variables de un proceso para el cual no existe la disponibilidad de un sensor físico. El procedimiento de diseño e implementación se ilustra mediante su aplicación a un intercambiador de calor de doble tubo con flujos a contracorriente. El enfoque empleado para diseñar el observador no lineal se basa en un modelo matemático simplificado del proceso. Se desarrollaron simulaciones numéricas y posteriormente pruebas experimentales en una planta piloto (intercambiador de calor) con el fin de validar el esquema propuesto.

En este artículo se presenta el desarrollo de un método de diagnóstico de fallas para sistemas no lineales representado por el enfoque de múltiples modelos. La principal contribución consiste en el diseño de un esquema de diagnóstico de fallas (FDD) a través de un filtro adaptativo desacoplado de fallas. El comportamiento del sistema está dado mediante múltiples modelos lineales alrededor de diferentes puntos de operación. Los residuos se establecen de tal manera que permiten estimar funciones de activación robustas a fallas. El filtro adaptivo diseñado permite la detección, localización y estimación de fallas múltiples o simultáneas. Las condiciones de estabilidad del filtro son estudiadas a través de LMI (Linear Matrix Inequality). El desempeño del esquema propuesto se valida en un sistema de tres tanques. In this article, a fault detection and diagnosis method is developed for nonlinear systems described by multi-models approach. The main contribution consists in the detection, isolation and estimation of faults for the nonlinear systems using a fault-decoupled adaptive filter. Based on the assumption that the dynamic behavior of the process is described by a multi-model approach around different operating points, a set of residuals is established in order to generate weighting functions robust to faults. These robust weighting functions are directly linked to the adaptive filter effectiveness which provides multiple fault magnitude estimations for the whole operating range of the system. Stability conditions of the adaptive filter are studied and its performance is tested using a three-tank hydraulic system.

En este artículo se presenta el desarrollo de un método de diagnóstico de fallas para sistemas no lineales representado por el enfoque de múltiples modelos. La principal contribución consiste en el diseño de un esquema de diagnóstico de fallas (FDD) a través de un filtro adaptativo desacoplado de fallas. El comportamiento del sistema está dado mediante múltiples modelos lineales alrededor de diferentes puntos de operación. Los residuos se establecen de tal manera que permiten estimar funciones de activación robustas a fallas. El filtro adaptivo diseñado permite la detección, localización y estimación de fallas múltiples o simultáneas. Las condiciones de estabilidad del filtro son estudiadas a través de LMI (Linear Matrix Inequality). El desempeño del esquema propuesto se valida en un sistema de tres tanques.

En este artículo se desarrolla un método para la representación de sistemas no lineales, así como un esquema de diagnóstico de fallas; en donde ambos utilizan un enfoque multi-modelos. El comportamiento dinámico del sistema se describe mediante el uso de diferentes puntos de operación por cada modelo lineal invariante en el tiempo. Por cada modelo un filtro de Kalman desacoplado de fallas es utilizado. Esto permite la generación de un banco de residuos con la finalidad de estimar funciones de ponderación robusta a fallas. Estas últimas se enlazan directamente al esquema de detección, localización y estimación de fallas con objeto de mejorar su eficacia. Finalmente, se realiza un ejemplo de simulación a fin de probar el desempeño del método desarrollado. In this article are developed a method for the representation of nonlinear systems as well as a scheme of fault diagnosis, both use a multi-model approach. The dynamic behavior of the system is described using different points of operation for every linear time invariant model. For every model a decoupled Kalman filter of faults is used. This allows the creation of a bank of residuals to estimate the robust weighting functions. This bank is directly linked to a scheme for detecting, isolating and estimating faults in order to improve its effectiveness. A simulation example to test the performance of the developed method is presented

In this paper we present a high-gain observer implemented in its discrete and continuous-discrete versions in order to compare the performance of both algorithms. The comparison is made considering the sampling time used to perform the observer’s correction stage in order to establish that the continuous-discrete observer is the best option when a low sampling time is used. Under this condition the continuous discrete observer can process data performing a reliable on-line estimation of the system (a slow dynamics of the process is required). We apply both algorithms to a distillation column that uses the Ethanol-Water binary mixture.