Sensitivity analysis for feedforward control system design

Abstract

A sensitivity analysis for feedforward-controlled finite domain systems is presented. Using an analytical approach, the sensitivity of a controlled system to variations in the design variables of the control inputs and the error sensors is obtained. Since the physical implementation of the control inputs acid the error sensors is directly related to the modal control forces and the modal error sensor components, the analysis is performed in the modal domain. This modal domain analysis offers the advantage that the highest computational effort in obtaining the sensitivities is independent of the physical characteristics of the transducers. From the results in the modal domain, the sensitivities of the controlled system with respect to particular physical parameters are obtained by using the chain rule of differentiation. Therefore, different types of transducers can be investigated within a minimum computational cost. The formulation can be easily incorporated into an optimization procedure for the optimal design of feedforward control systems. A numerical example in which the proposed formulation is compared to results obtained using finite differences is included. (C) 1995 Acoustical Society of America.