Fast location of process noise for nonlinear system identification

Erliang Zhang, Maarten Schoukens

Research output: Unpublished contribution to conferenceUnpublished paper

Abstract

The focus on system identification has been changing with more of an emphasis on nonlinear systems over recent years. Nonlinear system identification has been developed by focusing on specific classes of systems, such as block-oriented 1], NARMAX [2] and nonlinear state space systems [3]. The general noise framework which incorporates process noise is one of state-of-the-art challenges in nonlinear system identification. It is non-trivial to locate the process noise w.r.t. the static nonlinearity within the nonlinear system as the disturbance of the process noise is hidden in the measurement noise and nonlinear distortions under general excitations. Attempt has been made in our previous work 4] by developing a measurement protocol which uses a specially designed input that is periodic but non-stationary within one period. This technique is effective, however time-consuming as multiple measurements are needed. The present work contributes to locate the process noise for system modeling by proposing a fast experiment protocol using only one measurement and applying a simple sine signal or narrowband multisine signals. The amplitude of the (multisine)sine signal can be seen as the expected value of the non-stationary input [4].
Original languageEnglish
Pages34
Number of pages1
Publication statusPublished - 24 Apr 2017
EventWorkshop on Nonlinear System Identification Benchmarks 2017 - Vrije Universiteit Brussels, U-residence, Brussels, Belgium
Duration: 24 Apr 201726 Apr 2017
http://homepages.vub.ac.be/~mschouke/FILES/BenchmarkWorkshop2017_Abstracts.pdf

Conference

ConferenceWorkshop on Nonlinear System Identification Benchmarks 2017
CountryBelgium
CityBrussels
Period24/04/1726/04/17
Internet address

Fingerprint

Dive into the research topics of 'Fast location of process noise for nonlinear system identification'. Together they form a unique fingerprint.

Cite this