IMPROVED EIGENSTRUCTURE ASSIGNMENT CONTROLLER-DESIGN USING A SUBSTRUCTURE-BASED COORDINATE SYSTEM

A new control coordinate system based on the Craig-Bampton substructur e representation is examined for reduced-order-model-based controller design. Eigenstructure assignment controller design using the proposed modeling approach is shown to exhibit improved accuracy over conventi onal normal mode modeling in both closed-loop eigenvalue and eigenvect or attainment. The improvement is due to the coordinate system's abili ty to accurately predict actuator boundary conditions and, thus, actua tor command forces, when extensive model reduction has been employed. Two multi-input/multi-output design examples are used to illustrate th e practicality of the method and the resulting improved closed-loop be havior.