Fast Evolutionary Algorithm for Airfoil Design via Neural Network

Abstract

The main disadvantage of evolutionary algorithms for airfoil design problems is the high computational costs associated with the use of computational fluid dynamics solvers. This disadvantage was significantly reduced by using an existing technique which combines a genetic algorithm and a neural network to rapidly improve populations. In this technique, both genetic algorithms and a properly trained neural network search the design space, resulting in an interactive process between a genetic algorithm and a neural network that greatly improves the exploration power of the algorithm. Experimental results show that the implemented algorithm discovers the desired solution quickly and significantly reduces the overall costs of solving the problem with little or no reduction in robustness.