Use of canonical correlation analysis in component model reduction
Conference Paper
Overview
Additional Document Info
View All
Overview
abstract
This paper presents a method to reduce the model of each component in a large scale system, while taking into account the interactions with the rest of the system. The paper compares two methods: canonical correlation analysis and component cost analysis. First the interaction between component k and the rest of the system is described by canonical correlation analysis. (The state variables of component k are ranked according to the size of their canonical correlation coefficients). Next a component cost analysis is performed in the canonical correlation coordinates. (The state variables of component k are then ranked according to their component cost). Better performance properties are obtained by the component cost analysis method when it yields stability.