Recurrence Rate spectrograms for the classification of nonlinear and noisy signals

Hertrampf, T; Oberst, S

Recurrence Rate spectrograms for the classification of nonlinear and noisy signals

Hertrampf, T Oberst, S

Permalink

Publisher:: IOP Publishing Ltd
Publication Type:: Journal Article
Citation:: PHYSICA SCRIPTA, 2024, 99, (3)
Issue Date:: 2024-03-01

Open Access

Copyright Clearance Process

Recently Added
In Progress
Open Access

This item is open access.

Adobe PDF

Download Published versionAdobe PDF (893.36 kB)

View on publisher's site

View statistics

Full metadata record

Field	Value	Language
dc.contributor.author	Hertrampf, T
dc.contributor.author	Oberst, S https://orcid.org/0000-0002-1388-2749
dc.date.accessioned	2024-09-25T05:24:10Z
dc.date.available	2024-09-25T05:24:10Z
dc.date.issued	2024-03-01
dc.identifier.citation	PHYSICA SCRIPTA, 2024, 99, (3)
dc.identifier.issn	0031-8949
dc.identifier.issn	1402-4896
dc.identifier.uri	http://hdl.handle.net/10453/180974
dc.description.abstract	<jats:title>Abstract</jats:title> <jats:p>Time series analysis of real-world measurements is fundamental in natural sciences and engineering, and machine learning has been recently of great assistance especially for classification of signals and their understanding. Yet, the underlying system’s nonlinear response behaviour is often neglected. Recurrence Plot (RP) based Fourier-spectra constructed through <jats:italic>τ</jats:italic>-Recurrence Rate (<jats:italic>RR</jats:italic> <jats:sub> <jats:italic>τ</jats:italic> </jats:sub>) have shown the potential to reveal nonlinear traits otherwise hidden from conventional data processing. We report a so far disregarded eligibility for signal classification of nonlinear time series by training RESnet-50 on spectrogram images, which allows recurrence-spectra to outcompete conventional Fourier analysis. To exemplify its functioning, we employ a simple nonlinear physical flow of a continuous stirred tank reactor, able to exhibit exothermic, first order, irreversible, cubic autocatalytic chemical reactions, and a plethora of fast-slow dynamics. For dynamics with noise being ten times stronger than the signal, the classification accuracy was up to ≈ 75% compared to ≈ 17% for the periodogram. We show that an increase in entropy only detected by the <jats:italic>RR</jats:italic> <jats:sub> <jats:italic>τ</jats:italic> </jats:sub> allows differentiation. This shows that RP power spectra, combined with off-the-shelf machine learning techniques, have the potential to significantly improve the detection of nonlinear and noise contaminated signals.</jats:p>
dc.language	English
dc.publisher	IOP Publishing Ltd
dc.relation	http://purl.org/au-research/grants/arc/DP200100358
dc.relation	http://purl.org/au-research/grants/arc/LP200301196
dc.relation.ispartof	PHYSICA SCRIPTA
dc.relation.isbasedon	10.1088/1402-4896/ad1fbe
dc.rights	info:eu-repo/semantics/openAccess
dc.subject	01 Mathematical Sciences, 02 Physical Sciences
dc.subject.classification	General Physics
dc.subject.classification	49 Mathematical sciences
dc.subject.classification	51 Physical sciences
dc.title	Recurrence Rate spectrograms for the classification of nonlinear and noisy signals
dc.type	Journal Article
utslib.citation.volume	99
utslib.for	01 Mathematical Sciences
utslib.for	02 Physical Sciences
pubs.organisational-group	University of Technology Sydney
pubs.organisational-group	University of Technology Sydney/Faculty of Engineering and Information Technology
pubs.organisational-group	University of Technology Sydney/Faculty of Engineering and Information Technology/School of Mechanical and Mechatronic Engineering
pubs.organisational-group	University of Technology Sydney/UTS Groups
pubs.organisational-group	University of Technology Sydney/UTS Groups/Transport Research Centre (TRC)
pubs.organisational-group	University of Technology Sydney/UTS Groups/Centre for Audio, Acoustics and Vibration (CAAV)
pubs.organisational-group	University of Technology Sydney/UTS Groups/Transport Research Centre (TRC)/Associate Member
utslib.copyright.status	open_access	*
dc.rights.license	This work is licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0). To view a copy of this license, visit https://creativecommons.org/licenses/by/4.0/
dc.date.updated	2024-09-25T05:24:08Z
pubs.issue	3
pubs.publication-status	Accepted
pubs.volume	99
utslib.citation.issue	3

Abstract:

Abstract Time series analysis of real-world measurements is fundamental in natural sciences and engineering, and machine learning has been recently of great assistance especially for classification of signals and their understanding. Yet, the underlying system’s nonlinear response behaviour is often neglected. Recurrence Plot (RP) based Fourier-spectra constructed through τ-Recurrence Rate (RR τ ) have shown the potential to reveal nonlinear traits otherwise hidden from conventional data processing. We report a so far disregarded eligibility for signal classification of nonlinear time series by training RESnet-50 on spectrogram images, which allows recurrence-spectra to outcompete conventional Fourier analysis. To exemplify its functioning, we employ a simple nonlinear physical flow of a continuous stirred tank reactor, able to exhibit exothermic, first order, irreversible, cubic autocatalytic chemical reactions, and a plethora of fast-slow dynamics. For dynamics with noise being ten times stronger than the signal, the classification accuracy was up to ≈ 75% compared to ≈ 17% for the periodogram. We show that an increase in entropy only detected by the RR τ allows differentiation. This shows that RP power spectra, combined with off-the-shelf machine learning techniques, have the potential to significantly improve the detection of nonlinear and noise contaminated signals.

Please use this identifier to cite or link to this item:

http://hdl.handle.net/10453/180974