Showing papers by "Paul M. Frank published in 1999"

Advances in control : highlights of ECC '99

Abstract: Automatic Control - The Hidden Technology.- Asymptotic Dynamics in Adaptive Gain Control.- Control Problems in Multi-Modal Telepresence Systems.- A Perspective on Advanced Strategies for Process Control (Revisited).- Advanced Control for Hot Rolling Mills.- Fault-tolerant Control Systems.- Are there New Industrial Perspectives in the Control of Mechanical Systems?- Hybrid Systems: Modeling and Supervisory Control.- Nonlinear Descriptor Systems.- Two-Dimensional Linear Systems.- Verification of Logic Controllers for Continuous Plants.- Nonlinear Predictive Control and Moving Horizon Estimation - An Introductory Overview.

A novel design of structured observer-based residuals for FDI

Abstract: An approach for the design of structured residuals for fault detection and isolation is presented. Different to the more traditional observer schemes, the proposed approach requires only a reduced number of observers to the generation of a set of structured residuals. In the most favorable case only one observer is required and the detection and isolation of a set of faults can be done so like using a bank of observers. An interesting property of the proposed approach is the reduced online requirements for the implementation. The results are shown on an example.

Fault detection: Different strategies for modelling applied to the three tank benchmark — A case study

Abstract: In this contribution different modelling stragies for fault detection and isolation (FDI) in view of the model complexity are studied at the benchmark problem Three Tank System. The particularities of models needed for model-based FDI in contrast to models used in control are pointed out. Of special interest is the question of complexity. This depends basically on the given situation such as the kind of plant, the kind and number of faults to be detected, the demands for fault isolation and robustness and the measurements available. The paper shows for a certain process that diagnostic models for controllable and observable plants comprise only a partial description of the input/output model and are therefore less complex than those for control. This issue is discussed in terms of different model-based FDI approaches — analytical and data-based. The results are demonstrated in a case study, where several model-based FDI approaches are applied to the Three Tank System, which has been declared as a benchmark within the COSY project of the European Science Foundation (ESF).

Multiplicative fault isolation in linear systems

Abstract: An approach to fault isolation of multiplicative faults in linear systems is presented. After modeling the faults as multiplicative ones, a procedure for the design of structured residuals is proposed. The design is based on a modified version of a nonlinear disturbance decoupling algorithm used for nonlinear systems. Nonlinear techniques are required because of the time variant fault distribution matrix resulting of the multiplicative fault modeling. The contributions are the study of fault isolability, the relaxing of existence condition with respect to early results and the develop of an algorithm for the fault decoupling.

Genetic evolving of dynamic neural networks with application to process fault diagnosis

Abstract: The robustness issue in model-based diagnosis of process faults is addressed by means of artificial neural networks The symptoms are generated by using observer schemes with dynamic neural nets Their design is based on a hierarchical genetic algorithm, extended back-propagation method and multiobjective optimisation The evolutionary search of genetic type is used to find the optimal architecture of the dynamic networks Static networks are then used to classify the symptoms Application to a laboratory process illustrates the approach It refers to component and instrument fault detection and isolation in a three-tank system

Neural observer-based approach to fault-tolerant control of a three-tank system

Abstract: The simulated laboratory set-up Three-Tank System is investigated from the standpoint of fault-tolerant control. Different neural observer schemes are used for symptom generation. Symptom evaluation is based on neural classifiers. The output of the classification stage is provided to a supervisory unit. This switches on the appropriate remedy action in case of a process fault. For actuator and component faults, the reconfiguration strategies are based on the redundant hardware and optimised control laws. The reconfiguration in case of sensor faults is based on neural observers. The outputs of the latter replace the faulty sensor(s).

Application of fuzzy logic based methods of residual evaluation to the three tank benchmark

Abstract: This paper describes the application of fuzzy logic to the analysis of residuals generated using an observer scheme for the three tank benchmark. Two methods of residual analysis are investigated; fuzzy inference (i.e. rule based) and fuzzy clustering. The aim of both of the schemes is to simplify the task of the operator in determining the state of the plant (and any necessary action) by improving the presentation of the fault /no-fault information. Simulation results are presented for the three tank benchmark, and the two methods compared as to their effectiveness for residual evaluation in terms of isolation ability and ease of design.

Neural observer-based approach to fault detection and isolation of a three-tank system

Abstract: The simulated laboratory set-up Three-Tank System is investigated from the standpoint of fault-tolerant control. The problem of robust model-based diagnosis is therefore addressed. Dynamic neural networks with mixed structure are used to design different observer-based schemes. Symptom evaluation is based on static neural nets. They are used to classify the obtained residuals. Different classifiers and decision criteria are analysed. Experimental results of simulation are included into a comparative study. This refers to actuator, component and instrument fault detection and isolation.

Fault detection and isolation in nonlinear systems

Abstract: In this paper the problem of fault isolation in non linear systems is considered. Faults are formulated as parameter changes and a novel definition of fault detent ability is introduced. This connects a fault with its effect on the output of the system under consideration. Based on this definition, a practical way to test the fault detectability is introduced. In this context, a general result about fault isolation is presented. The result is based on the decoupling of faults in order to obtain a set of structured residuals. A procedure to calculate the required decoupling transformations is proposed. These results relax the existence conditions required by early approaches and represent a generalisation of the existing result for linear to non linear systems. The existence conditions are examined and a design procedure for the feasible case is proposed. Numerical examples are used to show the proposed ideas.

FDI based on parameter and output estimation: An integrated approach

Abstract: The pap or considors a proceduro for the fault isolation in linear systems. The proposed approach is based on the estimation of a reduced number of parameters of the original system. The selection of the parameters to be estimated follows from the evaluation of structured residuals. A reduction of the required input richness in the parameter estimation approach is reached while the fault isolability is conserved. Furthermore, an efficient way to combine the parameter and observer based approaches via parametrised observers Ls introduced. The proposed procedure is shown with an example.

Robust control for uncertain discrete-time jump linear systems with time delay

Abstract: This paper is concerned with the robust stochastic stabilization and H∞ disturbance attenuation for uncertain discrete-time Markovian jump linear time-delay systems subject to norm-bounded parameter uncertainties. The transition of the jumping parameters is governed by a finite-state Markov process. First a sufficient condition on stochastic stability is proposed using the stochastic Lyapunov functional approach. Then sufficient conditions on robust stochastic stabilizing and γ-suboptimal H∞ memoryless state feedback controllers are presented. It is shown that a stochastic stabilizing H∞ state feedback controller can be constructed through numerical solution of a set of coupled linear matrix inequalities.

Benchmark application of fuzzy observers to fault detection on a ship propulsion system

Abstract: This article is a contribution to the benchmark problem in the framework of the COSY (Control of Complex Systems) project and deals with fault detection on a simulated ship propulsion system. Our approach is to make use of a fuzzy output observer for the generation of the fault-reflecting residual signals. For this purpose, we first have to design suitable fuzzy models of the system under consideration. During the model-building stage, a lot of design parameters have to be chosen, such as for example the linguistic variables of the fuzzification and the composition operators in fuzzy space. On the other hand, using a fuzzy relational structure does not require the complete knowledge of all parameter values appearing in the system. So, we can take advantage of this fact during this design stage.

Estimation of wind gusts using an unknown input observer

Abstract: This paper describes the design of an estimator for wind gusts acting on an airplane during flight. The linearised longitudinal model of the airplane for steady state rectilinear flight is presented. It is shown, how wind gusts, which are considered to be unknown inputs to the airplane, can be estimated in general, if the unknown inputs appear in the output equation. The unknown inputs are removed from the system by use of a transformation. It is discussed, which measurements are necessary for the transformed system to be observable. The proposed estimator is a combination of an output transformation and an unknown input observer. The performance of the estimator is demonstrated with simulations. The robustness of the proposed estimator with respect to model uncertainties has also been tested with the help of simulations.

Tuning fuzzy observers for residual generation

Abstract: In this paper, we present an application study of a fuzzy output observer to fault detection on a laboratory three tanks system The concept used here is based on fuzzy relational model-building and model-based residual generation technique Several parameters are involved in the model-building procedure We focus our attention on their influence on the behaviour of the residuals We present the results obtained by the application of the observer to the detection of sensor failures The influence of the design parameters, such as the norm operators, the definition of linguistic variables or the choice of the identification algorithm is also discussed in this context

Fault isolation in a winding-machine using RCE networks

Abstract: The paper describes a mixed approach to the fault isolation on a winding machine. The first step is the generation of residuals, i.e. signals containing information about the faults. Based on a linear model of the winding machine, an observer based residual generation has been designed, however, measurements of the nonlinear system have been used for the calculation of the residuals. The second step is the residual evaluation and this is realised using Restricted Coloumb Energy Networks (RCE). The application of neural networks to this problems allows the isolation of different faults even if they are very small in comparison with the uncertainties. Note that the results achieved using the combined approach presented in this paper can not be achieved with known model based methods. It represents the modest contribution of this paper.

Modelling of a turbulence pickling line for adaptive control

Abstract: The paper presents a mathematical model of a turbulence pickling line, designer! to remow the oxide layer, known as scale, from a plain steel strip by means of a pickling medium, e.g. hydrochloric or sulphuric acid. The constituent parts of the installation and the differential equations describing the processes that take place inside of them, as well as effects encountered during the pickling process are overviewed. Further on, a simplified process model is described. Based on this, an adaptive control algorithm is developed. Simulation results and conclusions round up the presentation.

H ∞ detection filter design for state delayed linear systems with parameter uncertainty

Abstract: This paper studies the problem of Η ∞ detection filter design for state delayed linear continuous-time systems with parameter uncertainty, in which the effects of faults and disturbances cannot be de-coupled from each other. The detection filter gains are designed so that, if the fault of the system is absent, the filter is convergent for the admissible uncertainty conditions. The transfer function from disturbances to residual satisfies the prespecified constraint of H ∞ norm upper bound. Simultaneously, the design freedom can be used to achieve the extreme value of the residual-to-faults sensitivity.

Design of H ∞ detection filter with variance constraints and arbitrarily assigned eigenvalues

Abstract: Fault detection filters are a special class of observers that can generate directional residuals for the purpose of fault isolation This paper proposes a new approach to design H ∞ fault detection filters with variance constraints in which the eigenvalues can be arbitrarily assigned, so that the residual vector, generated by this filter, has directional properties This is done by combining the Η ∞ filter and fault detection filter principles The paper proposes a H ∞ fault detection filter and gives necessary and sufficient conditions for its existence After the H ∞ norm and the variance constraints are satisfied, the remaining design freedom can be used to make the residual have the directional property, based on the fault detection filter principle An illustrative example is given to show the applicability of the proposed approach