Spatiotemporal feature extraction using frequency-domain approach for flow-field data analysis
Open Access
- 1 December 2024
- journal article
- research article
- Published by AIP Publishing in Physics of Fluids
- Vol. 36 (12)
Abstract
Large-scale spatiotemporal data can easily be obtained from fluid-flow measurements owing to the development of measurement instruments. Data-driven analysis methods such as Proper Orthogonal Decomposition (POD) are used to analyze large-scale data and decompose flow-field structures into spatiotemporal correlations. In addition to decomposing the data into modes, selecting and extracting dominant features automatically are necessary. This study proposes a new analysis method that simultaneously selects the dominant POD modes and their corresponding dominant frequencies by estimating the amplitude and initial phase of the POD coefficients in the frequency domain rather than directly estimating the POD coefficients. The proposed method is applied to two types of experimental flow-field data to validate the method. In the first case, the proposed method is applied for noise reduction in pressure-sensitive paint data with a low signal-to-noise ratio. The results show that the proposed method can reduce noise more accurately and naturally than the other methods. In the second case, the dominant frequencies in the flow field and their spatial structures are estimated for vortex shedding at the trailing edge of the National Advisory Committee for Aeronautics 0012 airfoil. The data are acquired using particle image velocimetry. The results show that the proposed method can decompose flow-field data containing multiple frequency components into spatial correlations, and it can be used as a tool for feature extraction in the flow field.Keywords
Funding Information
- Institute of Fluid Science, Tohoku University (J24I064)
This publication has 36 references indexed in Scilit:
- Time-series image denoising of pressure-sensitive paint data by projected multivariate singular spectrum analysisExperiments in Fluids, 2022
- Quantitative evaluation of predictability of linear reduced-order model based on particle-image-velocimetry data of separated flow field around airfoilExperiments in Fluids, 2021
- Data-driven Classification and Modeling of Combustion Regimes in Detonation WavesFlow, Turbulence and Combustion, 2020
- Unsteady pressure-sensitive-paint (PSP) measurement in low-speed flow: characteristic mode decomposition and noise floor analysisExperiments in Fluids, 2019
- Classification of machine learning frameworks for data-driven thermal fluid modelsInternational Journal of Thermal Sciences, 2019
- Open-loop control of cavity noise using Proper Orthogonal Decomposition reduced-order modelComputers & Fluids, 2018
- De-biasing the dynamic mode decomposition for applied Koopman spectral analysis of noisy datasetsTheoretical and Computational Fluid Dynamics, 2017
- Fast PSP measurements of wall-pressure fluctuation in low-speed flows: improvements using proper orthogonal decompositionExperiments in Fluids, 2016
- Characterizing and correcting for the effect of sensor noise in the dynamic mode decompositionExperiments in Fluids, 2016
- Proper orthogonal decomposition closure models for turbulent flows: A numerical comparisonComputer Methods in Applied Mechanics and Engineering, 2012