Power spectrum estimation in innovation models

Abstract

In this paper, a regularized nuclear norm minimization-based subspace identification algorithm is proposed to estimate discrete-time power spectra of multi-input/multi-out systems in innovation models from output-only time-domain measurements. Parametrization and implementation issues are carefully addressed. The new identification algorithm and three popular subspace algorithms are tested on a numerical example and three real-life applications drawn from automotive engineering, acoustics, and renewable energy