Model Matching with Improved Transmission of Measuring Errors

This paper considers the problem of matching a system under control to a model using output feedback when both the given system and the model are described by nonparametric models, such as their impulse response matrices. The problem of matching is reduced to that of solving a linear system of equations by choosing a set of samples of the response impulses. The ``condition number´´ concept of numerical analysis is used in order to achieve a good transmission of measuring and/or computation errors towards the desired control results, which consist in this case in determining the feedforward and feedback compensator matrices necessary for matching purposes. If the distribution of the samples is adequate, the quality of the compensation may be greatly increased even under measuring (or computing) errors in observing (or calculating) the impulse response matrices. Applying this technique, one example is solved by simulation.