An Aeroelastic Metamodel Based on Experimental Data for Flutter Prediction of Swept Rectangular Wings

An aeroelastic metamodel was designed and implemented for prediction of flutter speed and frequency of swept rectangular wings based on experimental data and artificial neural networks (ANN). The ANN is a supervised multilayer perceptron that was trained based on an experimental data set involves flutter characteristics of various cantilever rectangular wing models. Some data were not learned to ANN and were maintained as test cases. The activation functions were tangent hyperbolic and linear function in the hidden and output layers respectively. For learning process, the normalized form of the inputs and outputs were given to the ANN. The ANN learned the relation between the inputs and outputs and was trained for predicting output parameters. It is observed that ANN results are in good agreement with experimental data as well as results of an aeroelasticity code developed using an analytical aerodynamic model. So this ANN can be used for quick prediction of flutter characteristics of swept rectangular wings and also for the study of the effects of various parameters on flutter characteristics of swept rectangular cantilevered wings.