Le Rle First Bezier

Transonic airfoil design problems are solved using a Genetic Algorithm (GA) based optimizer. At the desired operating point, the minimum drag and constant lift targets are achieved through either a scalarized objective function, involving an arbitrary weighting factor, or the Pareto technique. For the optimization of an airfoil at two operating points, similar approaches are used. The CPU cost of the optimizer is kept low through Artii-cial Intelligence. A m ultilayer perceptron is trained using already evaluated individuals and provides good, though approximate, tness predictions. With the regularly trained network, the direct ow s o l v er calls are noticeably reduced.