Model Order Reduction of via Improved LK Transformation and QK Reduction

摘要

This paper provides two approaches for model order reduction of large dynamical systems. The methods are based on an improved LK transformation and the Schur Decomposition. The improved LK method is based on applying a column pivoting technique for producing a QR factorization that would reveal the rank of a matrix. The QK reduction method exploits the Schur decomposition method to transform the state matrix into block diagonal form consisting of retained and residualized partitions. The methods exhibit robust numerical stability. A description of the LK transformation and a math-ematical description of the proposed techniques are given in the paper. The methods have been applied to a linear model of commercial transport aircraft. Numerical examples are given that show comparisons between full-order and reduced-order models.