Data-Driven Approaches in Gusty Aerodynamics: Insights from Sparse Surface Pressure Measurements
Loading...
Authors
Chen, Dashuai
Date
2024-04-11
Type
thesis
Language
eng
Keyword
data-driven , gust encountering , aerodynamic modeling , sparse sensor , surface pressure
Alternative Title
Abstract
Gusts are instantaneous and strongly unstable flows that aircraft frequently encounter. It is crucial to predict the sudden changes in gust loads accurately for the stable control of aircraft. However, modeling unsteady aerodynamic loads is a significant challenge in complex, gusty environments due to the associated complexities of flow separation and other nonlinearities.
This thesis employs data-driven approaches to accurately estimate gust loads from sparse surface pressure measurements. Additionally, the contribution of the sparse surface pressure taps to gust loads is evaluated, pointing out an optimal layout of the sensors. Furthermore, the low-dimensional dynamics and characteristics of unsteady surface pressure are studied on complex 2D and 3D flows, offering a new paradigm for aerodynamic state estimation and control.
Firstly, a nonlinear multilayer perceptron (MLP) is applied to estimate gust loads on a nonslender delta wing, demonstrating the model’s capability to capture the relationship between surface pressure and gust loads with minimum learning samples. The fluctuation of the dynamic response from the surface pressure measurements is then examined by a filtering process. Followed by a sensitivity analysis to evaluate the contribution of surface pressure taps to gust loads.
Subsequently, modal analysis is conducted on the unsteady surface pressure by utilizing linear proper orthogonal decomposition (POD) to identify patterns in low-dimensional space. The surface pressures of an SD7003 airfoil with pitching and plunging motions, and then on a more complex delta wing experiencing gusts, are accurately reconstructed by only three principal POD modes, which are found to be intrinsically related to mean flow structure, Reynolds number, and angle of attack.
Finally, a hybrid reduced-order deep learning-based model for gust load predictions is introduced for a nonslender delta wing. Drawing on the insights from modal analysis of surface pressure, the three principal POD modes are utilized as inputs of a deep learning model, combining long short-term memory (LSTM) and MLP approaches. The precise predictions of gust-induced lift and drag demonstrate the robustness and effectiveness of hybrid deep learning models in the field of highly unsteady aerodynamic load prediction.
Description
Citation
Publisher
License
Queen's University's Thesis/Dissertation Non-Exclusive License for Deposit to QSpace and Library and Archives Canada
ProQuest PhD and Master's Theses International Dissemination Agreement
Intellectual Property Guidelines at Queen's University
Copying and Preserving Your Thesis
This publication is made available by the authority of the copyright owner solely for the purpose of private study and research and may not be copied or reproduced except as permitted by the copyright laws without written authority from the copyright owner.
Attribution-NonCommercial-NoDerivatives 4.0 International
ProQuest PhD and Master's Theses International Dissemination Agreement
Intellectual Property Guidelines at Queen's University
Copying and Preserving Your Thesis
This publication is made available by the authority of the copyright owner solely for the purpose of private study and research and may not be copied or reproduced except as permitted by the copyright laws without written authority from the copyright owner.
Attribution-NonCommercial-NoDerivatives 4.0 International