kukuhiksanmusyahada/surrogate-model-using-dnn-and-pod

Aeroelastic Properties Prediction using Deep Learning with Proper Orthogonal Decomposition Method

Title

Development of Surrogate Model Using Deep Learning and Proper Orthogonal Decomposition

Problem Statements

• Predict the aerodynamic properties of airfoil NACA 64A010 using Surrogate Model
• The surrogate model will utilize Deep Learning as a model and Proper Orthogonal Decomposition as processing method
• The properties will be predicted are Lift Coefficient, Drag Coefficient, Plunging Motion, and Pitching Motion
• The model input should only be the flow characteristic (mach) and material characteristic (flutter speed index)
• The flight range will be mach 0.6 until mach 0.9
• Flutter speed range is vf 0.4 until vf 2.0

For the report and presentation can be accessed with this link

Results

Prediction Test Compared to Raw Data

Some of the prediction result is shown below. For more result can be seen from report or presentation report

Flutter Boundary Contour Plot

Model with training size 50 Samples

Model with training size 100 Samples

Model with training size 150 Samples

Flutter Boundary Result Compared to Previous Research