System identification of the hydrodynamic characteristics of underwater vehicles

Abstract

Accurate physics based simulations of Autonomous Underwater Vehicles (AUVs) and submarines require precise knowledge of the physical loads on the vehicle. Of these loads the hydrodynamic component is the most challenging to determine. One common method of representing these loads is to use hydrodynamic coefficients. These coefficients are commonly determined from captive model testing, which is expensive. However, in theory these coefficients can be found from free swimming trials. This thesis documents research undertaken, in collaboration with QinetiQ, to determine whether the hydrodynamic coefficients in a set of non-linear submarine equations can be determined from free swimming trials data.

Two coefficients identification procedures are described. The first, a non-linear approach attempts to find a set of coefficients reproduced the manoeuvring submarine’s path. The second, a linear approach recasts the identification task into a linear algebra problem which can be solved using standard techniques.

When testing using simulated data both approaches indicated that accurate submarine track identification is not equivalent to correct submarine hydrodynamic co-efficient identification. This arises from two causes.  First, it was discovered that the coefficients were not unique, and thus there are an infinite number of different coefficient sets which produce the same manoeuvre. Secondly, it was found that the manoeuvre always identified an ill-conditioned set of coefficients; that is small deviations in the manoeuvre produced large deviations in the identified coefficient values.

Due to these results, the results of the research suggest that it is not possible to determine the non-linear hydrodynamic coefficients of a submarine or AUV from free swimming manoeuvre data.

More information

Identifiers

Catalogue record

Export record

ASCII CitationAtomBibTeXData Cite XMLDublin CoreDublin CoreEP3 XMLEndNoteHTML CitationHTML CitationHTML ListJSONMETSMODSMPEG-21 DIDLOpenURL ContextObjectOpenURL ContextObject in SpanRDF+N-TriplesRDF+N3RDF+XMLRIOXX2 XMLReferReference ManagerSimple Metadata