Enhancing Readability of the Spectral Correlation Function using Multitaper Reassignment for Milling Vibration Feature Extraction
(2026) 10th Conference on High Performance Cutting, CIRP-HPC 2026 In Procedia CIRP 141. p.394-399- Abstract
Many vibration signals produced by rotating machines have been shown to be second-order cyclostationary. This means that the spectral correlation function can be used to identify cyclic frequencies associated with e.g. faulty machine parts. However, due to spectral leakage, the spectral correlation function often has poor localization of signal components, leading to sub-par fault detection. Recently, the Reassigned Spectral Correlation (RSC) method has been introduced to improve the localization of the spectral correlation function. This work proposes combining the RSC method with multitapering to achieve more noise-robust spectral correlation reassignment. Although the standard RSC method leads to much improved readability in... (More)
Many vibration signals produced by rotating machines have been shown to be second-order cyclostationary. This means that the spectral correlation function can be used to identify cyclic frequencies associated with e.g. faulty machine parts. However, due to spectral leakage, the spectral correlation function often has poor localization of signal components, leading to sub-par fault detection. Recently, the Reassigned Spectral Correlation (RSC) method has been introduced to improve the localization of the spectral correlation function. This work proposes combining the RSC method with multitapering to achieve more noise-robust spectral correlation reassignment. Although the standard RSC method leads to much improved readability in situations with little noise, the RSC method still has problems identifying signal components for noisier signals due to sensitivity of the reassignment coordinates. Multitaper techniques are well-studied in the signal processing literature and involve averaging over multiple spectral estimates and reassignment coordinates using mutually orthogonal window functions. The performance of the proposed multi-taper RSC is investigated using simulated cyclostationary signals. Although most of the previous cyclostationary literature applied to mechanical systems has been devoted to fault diagnosis, the method is shown to be able to detect cyclic features for experimental milling vibration signals where the cutting tools are undergoing progressive wear. This opens up for future applications of cyclostationary analysis in machining.
(Less)
- author
- Persson, Jonatan LU ; Åkesson, Maria LU ; Lindvall, Rebecka LU ; Sridhar, Gautam LU ; Gutnichenko, Oleksandr LU and Sandsten, Maria LU
- organization
-
- Sentio: Integrated Sensors and Adaptive Technology for Sustainable Products and Manufacturing
- Mathematical Statistics
- LU Profile Area: Light and Materials
- LTH Profile Area: Nanoscience and Semiconductor Technology
- NanoLund: Centre for Nanoscience
- SPI: Sustainable Production Initiative
- Production and Materials Engineering
- LU Profile Area: Natural and Artificial Cognition
- LTH Profile Area: AI and Digitalization
- LTH Profile Area: Engineering Health
- ELLIIT: the Linköping-Lund initiative on IT and mobile communication
- eSSENCE: The e-Science Collaboration
- publishing date
- 2026
- type
- Contribution to journal
- publication status
- published
- subject
- keywords
- Cyclostationary, Fault Diagnosis, Multitaper, Reassignment, Spectral Correlation, Tool Wear
- in
- Procedia CIRP
- volume
- 141
- pages
- 394 - 399
- publisher
- Elsevier
- conference name
- 10th Conference on High Performance Cutting, CIRP-HPC 2026
- conference location
- Cluny, France
- conference dates
- 2026-06-17 - 2026-06-19
- external identifiers
-
- scopus:105042608329
- ISSN
- 2212-8271
- DOI
- 10.1016/j.procir.2026.03.095
- language
- English
- LU publication?
- yes
- id
- ab80437e-6dfa-487e-88b7-8d7510594137
- date added to LUP
- 2026-09-21 15:39:30
- date last changed
- 2026-09-21 15:40:01
@article{ab80437e-6dfa-487e-88b7-8d7510594137,
abstract = {{<p>Many vibration signals produced by rotating machines have been shown to be second-order cyclostationary. This means that the spectral correlation function can be used to identify cyclic frequencies associated with e.g. faulty machine parts. However, due to spectral leakage, the spectral correlation function often has poor localization of signal components, leading to sub-par fault detection. Recently, the Reassigned Spectral Correlation (RSC) method has been introduced to improve the localization of the spectral correlation function. This work proposes combining the RSC method with multitapering to achieve more noise-robust spectral correlation reassignment. Although the standard RSC method leads to much improved readability in situations with little noise, the RSC method still has problems identifying signal components for noisier signals due to sensitivity of the reassignment coordinates. Multitaper techniques are well-studied in the signal processing literature and involve averaging over multiple spectral estimates and reassignment coordinates using mutually orthogonal window functions. The performance of the proposed multi-taper RSC is investigated using simulated cyclostationary signals. Although most of the previous cyclostationary literature applied to mechanical systems has been devoted to fault diagnosis, the method is shown to be able to detect cyclic features for experimental milling vibration signals where the cutting tools are undergoing progressive wear. This opens up for future applications of cyclostationary analysis in machining.</p>}},
author = {{Persson, Jonatan and Åkesson, Maria and Lindvall, Rebecka and Sridhar, Gautam and Gutnichenko, Oleksandr and Sandsten, Maria}},
issn = {{2212-8271}},
keywords = {{Cyclostationary; Fault Diagnosis; Multitaper; Reassignment; Spectral Correlation; Tool Wear}},
language = {{eng}},
pages = {{394--399}},
publisher = {{Elsevier}},
series = {{Procedia CIRP}},
title = {{Enhancing Readability of the Spectral Correlation Function using Multitaper Reassignment for Milling Vibration Feature Extraction}},
url = {{http://dx.doi.org/10.1016/j.procir.2026.03.095}},
doi = {{10.1016/j.procir.2026.03.095}},
volume = {{141}},
year = {{2026}},
}