Application of Artificial Neural Network based Surrogate Models for Parameterized Flow to a Sweptback Wing: An Outlook for Wind Turbine Blades

Date

2025-01-01

Author

Yilmaz, Ozge Ozkaya
Kayran, Altan

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

56
views

0
downloads

Cite This

The dynamics of blunt-body reentry vehicles are crucial, as they must withstand the intense aerodynamic forces and extreme heat generated during atmospheric reentry. Despite thermal protection benefits, the blunt body shape can cause dynamic instabilities, especially at transonic and low-supersonic speeds, which are crucial during parachute deployment and scientific measurements. Ensuring robust stability requires precise optimization of system parameters and initial conditions. Markov chain Monte Carlo (MCMC) framework, widely used for nonlinear parameter estimation for these systems, is highly sensitive to initial conditions. However, traditional gradient-based methods are prone to getting stuck in local minima, whereas neural network-based approaches often face challenges such as overfitting, hyperparameter sensitivity, and the curse of dimensionality in navigating complex loss landscapes. This study employs Differential Evolution (DE), an evolutionary algorithm, to optimize initial conditions for MCMC parameter estimation. We present a case study on 1-DoF planar motion blunt body vehicle dynamics, demonstrating that DE improves computational efficiency and solution quality. Using DE, the pitch-damping sum coefficients (a = 1.49, b =-0.78, c =-0.37, d =-0.31, and e =-1.01) were optimized to capture nonlinear aerodynamic behavior through a cubic spline model. The reconstructed angle-of-attack (a) trajectories closely matched high-fidelity CFD data, with a sum of squared errors of 700, validating the accuracy of approach. Additionally, DE reduced computational time significantly, achieving a 30.0% reduction when using 10 CPU cores and a 43.6% reduction with 15 CPU cores compared to 5 CPU cores computation. This research advances our understanding of blunt body dynamics and underscores the value of evolutionary algorithms in complex aerospace applications.

URI

https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=105001421762&origin=inward
https://hdl.handle.net/11511/114236

Collections

Department of Aerospace Engineering, Conference / Seminar