Jiawei Hu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Jiawei Hu has a total of 12 co-authors. Here are the top co-authors by number of shared publications:
- Hanru Liu
Associate Professor, School of Power and Energy, Northwestern Polytechnical Univ., Xi’an, Shaanxi 710129, People’s Republic of China; Associate Professor, National Key Laboratory of Science and Technology on Advanced Light-Duty Gas-Turbine, Xi’an, Shaanxi 710072, People’s Republic of China. ORCID:
2 shared publications Weiwei Zhang
2 shared publications- Weixiong Chen
Xi’an Jiaotong University
2 shared publications - Yangang Wang
School of Automation, Southeast University, China
2 shared publications - Hao Wang
Ministry of Education Key Laboratory of Micro and Nano Systems for Aerospace, Northwestern Polytechnical University, Xi'an, China
1 shared publication - Kuijun Zuo
Northwestern Polytechnical University, China
1 shared publication
Latest Publications
- Open Access
- 1 April 2024
- Published by AIP Publishing in Physics of Fluids
- Vol. 36 (4)
Fast fluid–structure interaction simulation method based on deep learning flow field modeling
The rapid acquisition of high-fidelity flow field information is of great significance for engineering applications such as multi-field coupling. Current research in flow field modeling predominantly focuses on low Reynolds numbers and periodic flows, exhibiting weak generalization capabilities and notable issues with temporal inferring error accumulation. Therefore, we establish a reduced order model (ROM) based on Convolutional Auto-Encoder (CAE) and Long Short-Term Memory (LSTM) neural network and propose an unsteady flow field modeling method for the airfoil with a high Reynolds number and strong nonlinear characteristics. The attention mechanism and weak physical constraints are integrated into the model architecture to improve the modeling accuracy. A broadband excitation training strategy is proposed to overcome the error accumulation problem of long-term inferring. With only a small amount of latent codes, the relative error of the flow field reconstructed by CAE can be less than 5‰. By training LSTM with broadband excitation signals, stable dynamic evolution can be achieved in the time dimension. CAE-LSTM can accurately predict the forced response and complex limit cycle behavior of the airfoil in a wide range of amplitude and frequency under subsonic/transonic conditions. The relative errors of predicted variables and integral force are less than 1%. The fluid–structure interaction framework is built by coupling the ROM and motion equations of the structure. CAE-LSTM predicts the time series response of pitch displacement and moment coefficient at different reduced frequencies, which is in good agreement with computational fluid dynamics, and the simulation time savings exceed one order of magnitude.
- 1 April 2024
- Published by AIP Publishing in Physics of Fluids
- Vol. 36 (4)
Open AccessFast fluid–structure interaction simulation method based on deep learning flow field modeling
The rapid acquisition of high-fidelity flow field information is of great significance for engineering applications such as multi-field coupling. Current research in flow field modeling predominantly focuses on low Reynolds numbers and periodic flows, exhibiting weak generalization capabilities and notable issues with temporal inferring error accumulation. Therefore, we establish a reduced order model (ROM) based on Convolutional Auto-Encoder (CAE) and Long Short-Term Memory (LSTM) neural network and propose an unsteady flow field modeling method for the airfoil with a high Reynolds number and strong nonlinear characteristics. The attention mechanism and weak physical constraints are integrated into the model architecture to improve the modeling accuracy. A broadband excitation training strategy is proposed to overcome the error accumulation problem of long-term inferring. With only a small amount of latent codes, the relative error of the flow field reconstructed by CAE can be less than 5‰. By training LSTM with broadband excitation signals, stable dynamic evolution can be achieved in the time dimension. CAE-LSTM can accurately predict the forced response and complex limit cycle behavior of the airfoil in a wide range of amplitude and frequency under subsonic/transonic conditions. The relative errors of predicted variables and integral force are less than 1%. The fluid–structure interaction framework is built by coupling the ROM and motion equations of the structure. CAE-LSTM predicts the time series response of pitch displacement and moment coefficient at different reduced frequencies, which is in good agreement with computational fluid dynamics, and the simulation time savings exceed one order of magnitude.
- Top Cited Papers
- 21 February 2023
- Published by Elsevier in Aerospace Science and Technology
- Vol. 136
Flow field modeling of airfoil based on convolutional neural networks from transform domain perspective
- 21 February 2023
- Published by Elsevier in Aerospace Science and Technology
- Vol. 136
Top Cited PapersFlow field modeling of airfoil based on convolutional neural networks from transform domain perspective
- 18 October 2022
- Published by Elsevier in Aerospace Science and Technology
- Vol. 130
Fast sparse flow field prediction around airfoils via multi-head perceptron based deep learning architecture
- 18 October 2022
- Published by Elsevier in Aerospace Science and Technology
- Vol. 130
Fast sparse flow field prediction around airfoils via multi-head perceptron based deep learning architecture
- Open Access
- 1 April 2021
- Published by Elsevier in Chinese Journal of Aeronautics
- Vol. 34 (4) , 341-351
Reduced order model for unsteady aerodynamic performance of compressor cascade based on recursive RBF
- 1 April 2021
- Published by Elsevier in Chinese Journal of Aeronautics
- Vol. 34 (4) , 341-351
Open AccessReduced order model for unsteady aerodynamic performance of compressor cascade based on recursive RBF
- Open Access
- 1 January 2020
- Published by Taylor & Francis in Engineering Applications of Computational Fluid Mechanics
- Vol. 14 (1) , 1501-1514
Sensitivity investigation of a subsonic cascade performance to geometric deviations based on statistical method
This paper has established an evaluation system for the sensitivity analysis of cascade/compressor aerodynamic performance to geometric deviations. The system includes blade profile parameterization with non-uniform rational B-spline (NURBS), uniform Latin hypercube sampling in association with Monte Carlo algorithm and statistical investigation methods. The sensitivity of a subsonic cascade performance to its geometric deviations is investigated under different typical inflow conditions. Results show that there is an approximately linear relationship between the geometric deviations and the aerodynamic parameters. The total pressure loss coefficient is the most sensitive to the leading edge profile deviation at i=2°. However the suction profile deviation and the leading edge profile deviation dominate the total pressure loss when the incidences are i = −4° and i = −8°. The blocking effect of the increased leading edge profile moves forward the separation point of the boundary layer. Therefore the scale of the separation vortex is enlarged, the total pressure loss is increased at i = 2° condition. Regarding i = −8°, the enlarged suction profile deviation reduces the circumferential static pressure gradient, resulting in an increase of the scale of separation vortex on the pressure side. The work is helpful to robust design of compressor blade profile and manufacturing & maintenance techniques of compressor blades.
- 1 January 2020
- Published by Taylor & Francis in Engineering Applications of Computational Fluid Mechanics
- Vol. 14 (1) , 1501-1514
Open AccessSensitivity investigation of a subsonic cascade performance to geometric deviations based on statistical method
This paper has established an evaluation system for the sensitivity analysis of cascade/compressor aerodynamic performance to geometric deviations. The system includes blade profile parameterization with non-uniform rational B-spline (NURBS), uniform Latin hypercube sampling in association with Monte Carlo algorithm and statistical investigation methods. The sensitivity of a subsonic cascade performance to its geometric deviations is investigated under different typical inflow conditions. Results show that there is an approximately linear relationship between the geometric deviations and the aerodynamic parameters. The total pressure loss coefficient is the most sensitive to the leading edge profile deviation at i=2°. However the suction profile deviation and the leading edge profile deviation dominate the total pressure loss when the incidences are i = −4° and i = −8°. The blocking effect of the increased leading edge profile moves forward the separation point of the boundary layer. Therefore the scale of the separation vortex is enlarged, the total pressure loss is increased at i = 2° condition. Regarding i = −8°, the enlarged suction profile deviation reduces the circumferential static pressure gradient, resulting in an increase of the scale of separation vortex on the pressure side. The work is helpful to robust design of compressor blade profile and manufacturing & maintenance techniques of compressor blades.