SOME PROPERTIES OF ROUTH CANONICAL REDUCED ORDER MODELS.

Routh canonical reduced-order models are very useful for evaluating suboptimal control policies for large-scale systems. The Routh Approximation Method (RAM) does not require a knowledge of system eigenvalues and eigenvectors and possesses many desirable features, such as preservation of stability and minimum computational requirements. The properties of the Routh canonical forms are investigated. The controllability, inversion of system matrices, approximate aggregation matrix, chained aggregation, and choice of model order aspects are included.