Decomposition of Large Scale Systems - A Numerical Approach

In the study of Large Scale Systems (LSS), one of the major concerns is the development of decomposition techniques. Decomposition of a large scale system into interconnection of lower-dimensional subsystems is instrumental to the analysis, estimation and control of LSS. While various decomposition methods have been developed in the past research, the computational aspects of the associated numerical algorithms are not yet fully explored. In this paper efficient and numerically reliable algorithms are proposed to decompose large scale systems with special emphasis upon the derivation of reduced order models. Decomposition and order reduction of the nearly singular and the time-scale separable systems are investigated by means of balancing transformations which can be computed via numerically reliable algorithms.