2024•Mechanical Systems and Signal ProcessingOpen access

Cepstral operational modal analysis for multiple-input systems based on the real cyclic cepstrum

Runyu Lu, Jérôme Antoni, Robert Bond Randall, Pietro Borghesani, Wade A. Smith, Zhongxiao Peng

Open full text 5 citations

Abstract

Operational modal analysis aims at identifying modal parameters of a system from response measurements only. Cepstrum-based operational modal analysis (OMA) has an advantage over conventional OMA methods, since it relaxes the assumption of white noise excitation and is able to retain the relative scaling between modes, which is important in many applications such as force identification. In previous research, the cepstrum-based OMA has been applied to reconstruct the forcing functions generated by a gear pair from response measurements, in order to obtain more direct diagnostic information. However, the cepstrum-based OMA can only cope with systems with a single excitation. To overcome this challenge, this paper proposes a novel technique, which can be applied to minimum-phase multiple input systems that have a cyclostationary excitation with a unique cyclic frequency. The technique first extracts the single-input system from the multiple-input case using the cyclic spectral density (CSD) of the measurement, equal to the spectral correlation density (SCD) evaluated at a particular cyclic frequency. The method then identifies the frequency response function (FRF) by fitting the real cyclic cepstrum of the measurement, this being the inverse Fourier transform of the log magnitude of the CSD. This method elaborates on a technique developed in previous research, which extracted the signal of interest from the complex cyclic cepstrum instead, this being the inverse Fourier transform of the logarithm of the CSD with unwrapped phase. In this paper, a limitation of the previously developed technique is identified and discussed for the first time, viz. problems with the phase unwrapping of the CSD. The technique in this paper overcomes the drawback in the previous research and exhibits better performance in noisy conditions. Both methods have been validated on synthetic signals as well as vibration measurements from a beam test rig. An attempt has also been made to apply them to vibration signals from a single-stage spur gear test rig. The method based on the real cyclic cepstrum is considered to have great potential to reconstruct the forcing functions from vibration signals measured from multi-stage gearboxes.

About this research paper

What this paper is about

Operational modal analysis aims at identifying modal parameters of a system from response measurements only. Cepstrum-based operational modal analysis (OMA) has an advantage over conventional OMA methods, since it relaxes the assumption of white noise excitation and is able to retain the relative scaling between modes, which is important in many applications such as force identification. In previous research, the cepstrum-based OMA has been applied to reconstruct the forcing functions generated by a gear pair from response measurements, in order to obtain more direct diagnostic information. However, the cepstrum-based OMA can only cope with systems with a single excitation. To overcome this challenge, this paper proposes a novel technique, which can be applied to minimum-phase multiple input systems that have a cyclostationary excitation with a unique cyclic frequency. The technique first extracts the single-input system from the multiple-input case using the cyclic spectral density (CSD) of the measurement, equal to the spectral correlation density (SCD) evaluated at a particular cyclic frequency. The method then identifies the frequency response function (FRF) by fitting the real cyclic cepstrum of the measurement, this being the inverse Fourier transform of the log magnitude of the CSD. This method elaborates on a technique developed in previous research, which extracted the signal of interest from the complex cyclic cepstrum instead, this being the inverse Fourier transform of the logarithm of the CSD with unwrapped phase. In this paper, a limitation of the previously developed technique is identified and discussed for the first time, viz. problems with the phase unwrapping of the CSD. The technique in this paper overcomes the drawback in the previous research and exhibits better performance in noisy conditions. Both methods have been validated on synthetic signals as well as vibration measurements from a beam test rig. An attempt has also been made to apply them to vibration signals from a single-stage spur gear test rig. The method based on the real cyclic cepstrum is considered to have great potential to reconstruct the forcing functions from vibration signals measured from multi-stage gearboxes.

Why it matters

OpenAlex reports 5 citations for this work. Citation counts describe recorded attention and do not establish research quality.

Key contribution

A contribution statement is not available in the OpenAlex record.

Method / approach

Method details are not available in the OpenAlex metadata.

Main findings

Findings are not separately available in the OpenAlex metadata.

Limitations

Limitations are not available in the OpenAlex metadata.

Applications

Application details are not available in the OpenAlex metadata.

Available abstract

Operational modal analysis aims at identifying modal parameters of a system from response measurements only. Cepstrum-based operational modal analysis (OMA) has an advantage over conventional OMA methods, since it relaxes the assumption of white noise excitation and is able to retain the relative scaling between modes, which is important in many applications such as force identification. In previous research, the cepstrum-based OMA has been applied to reconstruct the forcing functions generated by a gear pair from response measurements, in order to obtain more direct diagnostic information. However, the cepstrum-based OMA can only cope with systems with a single excitation. To overcome this challenge, this paper proposes a novel technique, which can be applied to minimum-phase multiple input systems that have a cyclostationary excitation with a unique cyclic frequency. The technique first extracts the single-input system from the multiple-input case using the cyclic spectral density (CSD) of the measurement, equal to the spectral correlation density (SCD) evaluated at a particular cyclic frequency. The method then identifies the frequency response function (FRF) by fitting the real cyclic cepstrum of the measurement, this being the inverse Fourier transform of the log magnitude of the CSD. This method elaborates on a technique developed in previous research, which extracted the signal of interest from the complex cyclic cepstrum instead, this being the inverse Fourier transform of the logarithm of the CSD with unwrapped phase. In this paper, a limitation of the previously developed technique is identified and discussed for the first time, viz. problems with the phase unwrapping of the CSD. The technique in this paper overcomes the drawback in the previous research and exhibits better performance in noisy conditions. Both methods have been validated on synthetic signals as well as vibration measurements from a beam test rig. An attempt has also been made to apply them to vibration signals from a single-stage spur gear test rig. The method based on the real cyclic cepstrum is considered to have great potential to reconstruct the forcing functions from vibration signals measured from multi-stage gearboxes.

Key concepts: Cepstrum, Operational Modal Analysis, Fourier transform, Frequency response, Cyclostationary process, Modal, Spectral density, Computer science

Related papers

Back to paper searchBrowse research topicsOriginal source
Cepstral operational modal analysis for multiple-input systems based on the real cyclic cepstrum — Research Paper | ScholarLens