Multivariable Identification of the MIT Interferometer Testbed

Abstract

A method of obtaining a uniquely identifiable, minimal, multi input multi output (MIMO) state space model is presented. The state space model is constructed from the matrix partial fraction expansion of the system transfer function matrix by using a dyadic decomposition of the matrix partial fraction expansion residue matrices. The advantages of using this approach are explored from the theoretical as well as the pragmatic point of view. The method is employed in the identification of a very lightly damped structure, a tetrahedral interferometer testbed truss, which is to be used for vibration control.