Minimum information loss method based on cross-gramian matrix for model reduction (CGMIL)

Model reduction was studied by adopting the information theoretic principles and methods. A definition of system ldquocross-Gramian informationrdquo (CGI) was defined based on the information theoretic properties of the system cross-Gramian matrix. By analyzing the information descriptions of system states, the physical meanings of CGI was clarified from entropic points of view. The minimum loss of CGI was our performance index of model reduction. A new improved minimum information loss (MIL) methods denoted by CGMIL was developed for model reduction. As the theoretical analysis and simulation indicated, CGI is a good information description and CGMIL algorithm is better than MIL algorithm in the performance of model reduction.