Subspace-Based Fractional Order Systems Identification in Presence of Outliers
6 Pages Posted: 9 Nov 2018
Date Written: July 2018
This paper aims at identifying the commensurate fractional order systems when output signal is corrupted by noise and outliers. A new outliers detection method via nuclear norm and infinite norm is proposed, which can detect outliers exactly and estimate noise at the same time. With the adoption of recovered “clean” output signal, the multivariable output error state space method is used to estimate system parameters. The effectiveness and superiority of proposed method are verified by detailed numerical examples.
Suggested Citation: Suggested Citation