Estimating the spectral abscissa for switched linear systems via square coordinate transformations.

Switched linear systems are important in providing a convenient method to represent many practical systems. This paper investigates the performance analysis of switched linear systems under arbitrary switching. Computational algorithms based on square coordinate transformations are presented to obtain the least μ1 measure. Theoretical results showed that sequences of the least μ1 matrix set measure obtained by two different algorithms are convergent and the limit gives an upper bound estimation of the spectral abscissa. A numerical example is presented to show the effectiveness of the proposed method. [ABSTRACT FROM AUTHOR]