Jean-François Magni

A generalized approach to observers for fault diagnosis

A procedure for observer synthesis by eigenstructure assignment is described. The proposed procedure gives a unifying framework for all kinds of observers. The parity space approach is interpreted within this context. The main objective is to give a necessary and sufficient condition for the existence of residuals decoupled from a given set of failures or false alarm causes. The condition appears to be the same in the observer and in the generalized parity space cases.<<ETX>>

A robust self-scheduled missile autopilot: design by multi-model eigenstructure assignment

Abstract This paper describes the synthesis of a self-scheduled controller robust with respect to parametric uncertainties. Traditionally, scheduling is done a posteriori by interpolation of gains computed for several points within the parameter space. The method proposed here, named (Mu-μ)-iteration, is based on worst-case analysis and multi-model eigenstructure assignment. It permits the designer to synthesize a robust low order controller by considering a priori its dependence of some scheduling parameters. The method is then applied to derive a robust missile autopilot.

A skew mu toolbox (SMT) for robustness analysis

The aim of this freeware is to provide computational (skew) mu methods for analysing the robust stability and performance properties of an uncertain closed loop. It could also be considered as a software complement to the book. The toolbox contains low-level routines as well as fully automated procedures which allow a non specialist to obtain guaranteed robustness margins. Different realistic engineering applications are included (missile, rigid and flexible aircraft, telescope mock-up), which illustrate the efficiency and reliability of the proposed tools

Robust flight control design with respect to delays, control efficiencies and flexible modes

Abstract This paper presents an iterative technique which makes it possible to design a robust flight control system. The methodology described in this paper has multiple applications. Two different applications are considered: the first consists of computing a robust feedback with respect to parameter uncertainties by considering a bank of models, while the second provides a robust flight-control design for a highly flexible aircraft with significant cross-coupling between flight mechanics and structural dynamics modes. Theproposed method will solve these problems via an iterative optimization under linear inequality constraints. Finally, the iterative algorithm is applied to a civil aircraft control system in lateral flight.

Mixed-μ-Analysis for Flexible Systems. Part II: Application to the Spot Family of Earth Observation Satellites

Abstract Mixed-μ analysis of the companion paper (Magni et al., 1999) is a new tool introduced to support the control system design process on complex, multivariable flexible dynamic systems. In the present paper the transfer from research to application is described. The example retained is the SPOT satellite developed at MMS for earth imagery. The specificities of flexible large Space structures are correctly accounted for by the new tools, as they offer the capacity to track very narrow resonant peaks. All known as well as nonintuitive though meaningful worst-case combinations were discovered

An LFT approach to robust gain scheduling

Traditionally, scheduled controllers are obtained by interpolation of a bank of linear feedback gains. The strategy adopted here is quite different as the controller is directly designed in scheduled form. The key idea consists of designing the feedback gain in LFT (Linear Fractional Transformation) form, the system to be controlled being itself in LFT form. Using an observer-based scheduled feedback and the corresponding Q-parameterization, robust control design can be treated by alternating μ-analysis for worst case identification and multimodel control for simultaneous treatment of the worst cases.

Enhanced LFR-toolbox for MATLAB

We describe recent developments and enhancements of the LFR-toolbox for MATLAB for building LFT-based uncertainty models. A major development is the new LFT-object definition supporting a large class of uncertainty descriptions: continuous- and discrete-time uncertain models, regular and singular parametric expressions, more general uncertainty blocks (nonlinear, time-varying, etc.). By associating names to uncertainty blocks the reusability of generated LFT-models and the user friendliness of manipulation of LFR-descriptions have been highly increased. Significant enhancements of the computational efficiency and of numerical accuracy have been achieved by employing efficient and numerically robust FORTRAN implementations of order reduction tools via Mex-function interfaces. The new enhancements in conjunction with improved symbolical preprocessing lead generally to a faster generation of LFT-models with significantly lower orders

A general framework for pole assignment algorithms

The problem of pole assignment by gain output feedback or by low-order dynamical compensator is considered from a geometrical point of view. This allows unification of a general framework for most of the existing pole assignment methods formulated in a state-space context, such as the minimal-order observers, the F.M. Brasch and J.B. Pearson (1970) compensator, the methods proposed by H. Kumura (1978), etc., and simplification of their presentation. Pole assignment algorithms are derived from this general formulation.<<ETX>>

Flexible transmission system controlled by modal dynamic feedback

This paper presents a method for eigenstructure assignment by dynamic feedback. This method can be used considering either single input single output or general multivariable systems. Robustness is taken into account in a multimodel setting. The design problem considered is shown to reduce to solving a set of linear equations. A quadratic criterion which enables to deal with performances and sensitivity functions is also proposed. The proposed method is illustrated considering a Flexible Transmission System benchmark problem.

Mixed-μ-Analysis for Flexible Systems. Part I: Theory 1

Abstract This paper presents new tools for computing upper and lower bounds of μ without frequency gridding. The proposed techniques for computing lower bounds of the peaks of the μ-curve, are divided into two steps. The first one consists of finding the perturbation with minimum Frobenius norm that leads to the limit of stability. Using this result as an initialization, the second algorithm finds the perturbation with minimum sigma-max norm such that the system remains at the limit of stability. The limit of stability is considered both from a state space and from a transfer matrix point of view, which leads to two classes of techniques. The lower bounds are validated by using an upper-bound analysis technique that considers standard scalings over a range of frequencies instead of at an isolated frequency