Akademik Veri Yönetim Sistemi
AIAA Aviation Forum and ASCEND, 2024, Nevada, Amerika Birleşik Devletleri, 29 Temmuz - 02 Ağustos 2024
Data acquired from computationally expensive studies need effective post processing strategies for the assessment of the models with different variables. For this purpose, this study presents high-fidelity surrogate models constructed for the prediction of Cp distribution exploiting the ensemble of data resulting from the static aeroelastic analyses of a three-dimensional sweptback wing. Radial Basis Function (RBF) and Artificial Neural Network (ANN) methods are exploited to create a surrogate model and Cp distribution for different flow parameters that do not exist in the data set used for the construction of surrogate models, are predicted utilizing these models. As a by-product of these surrogate models, prediction of the deflection of the wing is also be made. The results of the methods presented are compared in terms of accuracy and computational efficiency. Moreover, it is shown that utilizing the Proper Orthogonal Decomposition (POD) method, number of outputs fed to the ANN can be greatly reduced which increases the efficiency of ANN based surrogate model; the computational cost is decreased drastically and accuracy increases.