Using singular values for model reduction of flexible structures

Finite-element methods (FEMs) and extraction from experimental data are the most widely used methods for developing models of flexible structures for the analysis, design, and simulation of control systems. However, these methods produce models with modes that do not contribute significantly to the fidelity of the model. Eliminating these modes is the major focus of this work. Particular attention is paid to MIMO (multiple-input multiple-output) systems. Control system model development, using finite-element methods and experimental data, is briefly discussed. The problems associated with models, including unnecessary and/or residual modes, are delineated. SISO (single-input, single-output) and MIMO techniques for reducing the order of models are presented, and example modal reduction is illustrated using modal data from the proposed NASA Shuttle-C.<>