Exposing resonances using the Bootstrapped Total Least Squares estimator

Dries Peumans, Antonin De Vestel, Cedric Busschots, Yves Rolain, Rik Pintelon, Gerd Vandersteen

Research output: Unpublished contribution to conferencePoster

29 Downloads (Pure)

Abstract

Almost all mechanical systems are in one way or another susceptible to resonant behaviour. This resonant behaviour should be accurately characterised to prevent unwanted mechanical oscillations which could in turn lead to mechanical failure. It is therefore of utmost importance to accurately characterise these mechanical resonances. In the past, advanced local rational modelling techniques have been introduced which effectively estimate the non-parametric Frequency Response Function (FRF) for Multiple-Input, Multiple-Output (MIMO) systems. Unfortunately, most of these existing techniques derive the local rational model parameters using a biased estimator. We developed a local rational modelling technique which removes the bias from the estimation procedure and is also applicable to MIMO systems. At its core, this technique uses the Bootstrapped Total Least Squares (BTLS) estimator, which not only provides a consistent estimate of the common-denominator rational model, but also allows us to generate accurate uncertainty bounds on the FRF matrix. The performance of the proposed technique was verified by characterising the resonant behaviour of the tailplane of a glider.
Original languageEnglish
Publication statusPublished - 24 Sep 2018
Event2018 ERNSI Workshop on System Identification - Cambridge, United Kingdom
Duration: 23 Sep 201826 Sep 2018

Workshop

Workshop2018 ERNSI Workshop on System Identification
Country/TerritoryUnited Kingdom
CityCambridge
Period23/09/1826/09/18

Fingerprint

Dive into the research topics of 'Exposing resonances using the Bootstrapped Total Least Squares estimator'. Together they form a unique fingerprint.

Cite this