Control-Relevant Parameter Estimation: A Systematic Procedure for Prefilter Design

The objective of this paper is to develop a systematic procedure for deriving control-relevant parameter estimation algorithms for linear models represented via the prediction-error model structure. The key element in the design procedure is the prefiltering of the input and output time series obtained from the plant. The prefiltering step insures that the estimated model retains those plant characteristics that are most significant with regards to the user's control requirements. In this paper we employ linear fractional representations of the closed-loop system to obtain a general statement of the control-relevant parameter estimation problem which applies to different types of models and control structures. The prefilters obtained via this technique incorporate explicitly the model structure, the desired closed-loop transfer functions, and the setpoint/disturbance characteristics of the control problem. The proposed procedure is then applied to obtain prefilters for models to be used to design feedback, feedforward, and decentralized controllers.