Passivity enforcement via spectral factorization
Please login to view abstract download link
We present a novel method for passivation of linear time-invariant dynamical systems. The method is based on the spectral factorization of the Popov function. In more detail, we fit a spectral factor to frequency data of the system. This is done by solving a constrained linear system of equations. From the spectral factor, we construct a passive realization of the system by modifying the output matrix and, optionally, the feedthrough matrix. The method can be easily extended to identifying a passive system from frequency data by combining it with, for example, the Loewner framework or vector fitting. We present several numerical examples to demonstrate the power of the method.
