Search

The problem of fault detection and isolation in a class of nonlinear systems having a Hamiltonian representation is considered. In particular, a model of a planar vertical take-off and landing aircraft with sensor and actuator faults is studied. A Hamiltonian representation is derived from an Euler-Lagrange representation of the system model considered. In this form, nonlinear decoupling is applied in order to obtain subsystems with (as much as possible) specific fault sensitivity properties. The resulting decoupled subsystem is represented as a Hamiltonian system and observer-based residual generators are designed. The results are presented through simulations to show the effectiveness of the proposed approach.

Robust and early detection of oscillatory failures in the electrical flight control system of an aircraft is a crucial issue. Oscillatory failures, if not detected and passivated in time, can lead to strong interactions with loads and aeroelasticity and may potentially lead to structural damages. A nonlinear observer-based solution to detect such failures with small amplitude at a very early stage is presented. The stability and convergence proofs are given and experimental results with real A380 flight data are presented.

In this paper, we address the problem of Oscillatory failure Case (OFC) detection in A380 aircraft. The importance of OFC detection is well known in the literature because of natural mode excitation and the structural loads they can provoke. A novel nonlinear observer-based approach, which has two different gains is proposed in this paper to solve the problem. The main advantage of the proposed residual generator is that, by an adequate selection of observer gains, the residual generator has a kind selective (frequency dependent) attenuation property: disturbances and unmodelled dynamics effects on the residual are reduced while the effect of the fault is slightly amplified in a specific fault frequency range. Further, if the gains are chosen to be zero, the particular case of the technique implemented on A380 Airbus is obtained. A test based on real measurements of an A380 airplane is presented to show the efficiency of the proposed residual over the existing techniques.

Internal transformer faults could cause severe damage to electrical networks and also deteriorate the energy quality standard. Most of existing approaches for fault detection in transformers are based on signal processing methods or a static parity check of functionality. Two analytical redundancy approaches one based on dynamic parity checs and a second using a receding horizon observer, to internal fault detection in transformers are proposed in this paper. The main advantage of our proposal is that explicit knowledge of the Lipschitz constant of the transformer nonlinearities is not required and the considered algorithms has a high degree of portability. Furthermore, the effect of load and external faults on the residual are decoupled. Stability properties of the model are demonstrated. Some simulations shown the efficiency of the proposed detection method and highlight the reliability of the considered approach.

Abstract This work is concerning with the problem of determinate if a fault has been originated in a subsystem of a electric network or in somewhere else. This problem is motivated for the legal aspect related to co-generation systems. To solve the problem an approach based on analytical redundancy is proposed. An electrical part of a co-generation system is used to show (via simulations) the success of the proposed approach.

This paper considers an observer-based approach for detection and isolation of actuator and component faults in a Fluid Catalytic Cracking (FCC) unit. Model based analytical redundancy is used on an augmented linearised model of the non-linear FCC unit to the isolation of faults. Some component faults are modelled and implemented in a software package to show the considered approach. The proposed design is tested using a non-linear simulation of the FCC unit model. The results show that the linear observer-based approach applied to an augmented state of the FCC to actuator and component fault detection and isolation can be used with success even if the non-linear behaviour of the FCC has some effect on the residuals.

This work is concerned with the design of a fault detection and isolation (FDI) system to monitor malfunctions in sensors and actuators of a fluid catalytic cracking (FCC) unit model predictive control (MPC) system. The control system is based on an infinite-horizon MPC algorithm. The fault detection scheme is developed based on two banks of robust observers, while the fault isolation task is done employing a structured residual approach. The models used in the control and in the FDI of the FCC unit are obtained by using subspace identification methods. The effectiveness of the proposed strategy is verified through numerical simulations carried out on a dynamical model of an industrial FCC unit under abrupt fault in its control devices.

Este artículo presenta un método de control tolerante a fallas para sistemas no lineales planos. Las propiedades intrínsecas d e los sistemas planos generan redundancia analítica y permiten calcular todos los estados y las entradas de control del sistema. Los residuo s son calculados comparando las medidas reales provenientes de los se nsores y las señales obtenidas gracias al conjunto de ecuacione s del sistema plano. Fallas multiplicativas y aditivas se pueden man ejar de manera indistinta. Las señales redundantes obtenidas co n las ecuaciones del sistema plano son usadas para reconfigurar el si stema con falla. La factibilidad del método propuesto es verifi cada para fallas aditivas en un sistema de tres tanques.

Fault tolerant control (FTC) seeks to avoid the deterioration of the system at the time that there is a change of one or more parameters outside the limits of tolerance. There is in the literature a set of approaches based on the idea of tracking the nominal trajectory, which have been proved to be effective. However, some of them do not maintain the nominal controller, others, even if they keep it, require meet very demanding existence conditions. This paper proposes a fault recovery using a cascade control structure. The proposed design is based on the nominal trajectory tracking as well, but keeps intact the original control while existence conditions are relaxed. The resulting equivalent controller can be changed according to the performance needs of the system's response to faults. The controller could be defined by using stability regions and LMI tools.

This paper presents a fatness based active fault tolerant control technique. This approach is based on flat systems theory and on the non-uniqueness property of these to flat outputs. This property allows to develop an accurate fault detection and isolation technique, in addition, it allows to generate non-faulty analytically redundant measures. These signals can be used to reconfgure the system after a fault. This approach is applied to a three tank system.

This paper addresses a fault detection and isolation technique for differential flat systems. For such nonlinear systems, it is possible to find a set of variables, named flat outputs such that states and control inputs can be expressed as functions of flat outputs and their derivatives. Flat systems properties are used to detect and isolate faults and the non-uniqueness property of the set of flat outputs is used for increase the number of residues and improve the fault isolation, the proposed approach will be applied on a classical three tank system.

A novel non-linear observer for detection of oscillatory faults in commercial airplanes is proposed in this work. Oscillatory Failure detection in airplanes is an important problem because of natural mode excitation and the structural loads they can provoke. A simple and novel non-linear observer with two correction terms is developed. Formal proof of stability and convergence of the observer in the fault free case is provided. The proposed approach satisfies technical requirements of no false alarm, small detection time, detectability of small faults while keeping a simple structure. A test with real measurements is presented to show the efficiency of the proposed observer over the existing techniques.