Optimal Feed-Forward Digital Tracking Controller Design

Abstract

This paper presents an approach for optimal digital feed-forward tracking controller design. The tracking problem is formulated as a model matching problem, in which the distance between a specified tracking reference model and the achievable tracking performance by feedforward compensation is minimized. Desired input/output characteristics, finite length preview action, tracking of specific classes of constrained signals, time domain reference signal velocity or acceleration bound, and frequency domain weighting are conveniently incorporated in the proposed controller design and their roles in tracking performance are discussed. The tracking error bound is also explicitly expressed in terms of the controller design parameters. An l1 norm optimal tracking controller is proposed as a solution to the mechanical tolerance control problem. A motion control example illustrates the design approach and several aspects of the resulting optimal feedforward controller, including the optimality of the zero phase error tracking controller.