Showing papers by "Ron J. Patton published in 1996"

Design of unknown input observers and robust fault detection filters

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 robust (in the disturbance de-coupling sense) fault detection filters which ensure that the residual vector, generated by this filter, has both robust and directional properties. This is done by combining the unknown input observer and fault detection filter principles. The paper proposes a new full-order unknown input observer, and gives necessary and sufficient conditions for its existence. After the disturbance de-coupling conditions are satisfied, the remaining design freedom can be used to make the residual have the directional property, based on the fault detection filter principle. A nonlinear jet engine system is used to illustrate the robust fault isolation approach presented. It is shown that linearization errors can be approximately treated as unknown disturbances and be de-coupled in the design of a robust fault detect...

Optimal filtering and robust fault diagnosis of stochastic systems with unknown disturbances

Abstract: The paper studies the optimal filtering and robust fault diagnosis problems for stochastic systems with unknown disturbances. An optimal observer is proposed, which can produce disturbance decoupled state estimation with minimum variance for time varying systems with both noise and unknown disturbances. The existence conditions and the observer design procedure are presented. The output estimation error with disturbance decoupling and minimum variance properties is used as a residual signal. A statistical testing procedure is applied to examine the residual and is used to diagnose faults. The method developed is applied to an illustrative example, and simulation results show that the optimal observer can give good state estimation; the fault detection approach taken is able to detect faults reliably in the presence of both modelling errors and noise.

Optimal residual design for fault diagnosis using multi-objective optimization and genetic algorithms

Abstract: This paper develops a new approach to the design of optimal residuals in order to diagnose incipient faults based on multi-objective optimization and genetic algorithms. In this approach the residual is generated via an observer. To reduce false and missed alarm rates in fault diagnosis, a number of performance indices are introduced into the observer design. Some performance indices are expressed in the frequency domain to take account of the frequency distributions of faults, noise and modelling uncertainties. All objectives then are reformulated into a set of inequality constraints on performance indices. A genetic algorithm is thus used to search for an optimal solution to satisfy these inequality constraints on performance indices. The approach developed is applied to a flight control system example, and simulation results show that incipient sensor faults can be detected reliably in the presence of modelling uncertainty.

Torque and flux estimation for a rail traction system in the presence of intermittent sensor faults

Abstract: A model based approach to making an induction motor drive tolerant to intermittent disconnections of the current and voltage sensors is considered. Whilst these are abrupt faults they are hard to detect due to their short duration. It is shown that a simple parity equation method is an obvious approach to diagnosing sensor faults, however the application of model based techniques is actually preferable for achieving reliable fault detection and isolation (FDI) when faced with intermittent faults. The induction motor dynamics have a strongly bi-linear structure. A novel bilinear observer is developed, then extended to produce a scheme for detecting and correcting sensor faults.

An application of eigenstructure assignment to robust residual design for FDI

Abstract: The eigenstructure assignment approach to fault detection and isolation (FDI) has been available for a number of years and now industrial applications are timely. This paper describes the application of this technique to the generation of robust residuals for FDI in a nuclear reactor coolant pumping system. The research is being evaluated through a real application project at the Heysham 2 power station (Nuclear Electric) and AEA technology (Risley).

Fault diagnosis using quantitative and qualitative knowledge integration

Abstract: This paper presents a novel approach to integrating quantitative and qualitative information in fault-diagnosis, and which is based on the use of associative B-spline functions. The underlying concept is to structure an artificial neural network which can model highly nonlinear systems efficiently, in a fuzzy logic format. The network could therefore be trained more rapidly and will also provide a linguistic description about the causes of faults. The diagnosis approach is put to the test through a digital simulation study of a nonlinear two-tank system.

Robust parametric eigenstructure assignment

Abstract: AIAA, Guidance, Navigation and Control Conference, San Diego, CA, July 29-31, 1996 This paper is concerned with robust eigenstructure assignment for multivariable systems. It combines time-domain performance specifications provided by eigenstructure assignment and robust performance specifications in the frequency domain considered by H-infinity control to realize joint optimal robust control design. A parametric expression for state-feedback eigenstructure assignment is derived for the case where the sets of closed- and open-loop eigenvalues do not intersect, which is based on a set of free parameters. All complex operations are converted into the real field so that the algorithm which is developed for the controller design can be easily implemented on computers. It uses a robustness index defined in the frequency domain as the cost function. The analytical gradient calculation of the cost function with respect to the free parameters is given. Using gradient-based optimization, the robustness index is minimized by making full use of the freedom provided by eigenstructure assignment. (Author)

Automated fault diagnosis using neuro-fuzzy systems

Abstract: This paper presents a novel approach to integrating quantitative and qualitative information in faultdiagnosis, based on the use of neuro-fuzzy systems. In this approach the residuals are generated and evaluated via a B-Spline functions network. The configuration adopted allows the designer to both extract and include symbolic knowledge from the trained network. The diagnosis approach is put to the test through a digital simulation study of a non-linear two-tank system.

On Eigenvectors and Generalized Eigenvectors in Multi-Variable Descriptor Control Systems

Abstract: This paper is concerned with an eigenvector problem in multi-variable descriptor control systems. It proves that the eigenvectors and generalized eigenvectors premultiplied by the singular matrix E of the descriptor system as well as the eigenvectors and the generalized eigenvectors, which play a very important role in the design of multi-variable descriptor systems via eigenstructure assignment, are linearly independent.