Aerodynamic inverse optimisation problems
Shigeru Obayashi · Institution of Engineering and Technology eBooks · 1997
Characteristics of aerodynamic optimisation have been discussed through wing shape design problems. It has been demonstrated that distribution of the objective function can be extremely rough even in a simplified problem. In such a situation, GA (genetic algorithm) is expected to be more effective than a simple hill-climbing strategy. Three optimisation algorithms, the GM (gradient-based method), SA (simulated annealing) and GA, were first applied to the airfoil shape design using the approximation concept to compare their performances. Although GA is time consuming, its result is superior to those of the others. Since the other algorithms will require many trials starting from various initial designs to obtain a comparable result, they will not have any advantage in efficiency. The result suggests that GA is the best option for aerodynamic optimisation.