Designing airfoils using a reference point based evolutionary many-objective particle swarm optimization algorithm

In this paper, we illustrate the use of a reference point based many-objective particle swarm optimization algorithm to optimize low-speed airfoil aerodynamic designs. Our framework combines a flexible airfoil parameterization scheme and a computational flow solver in the evaluation of particles. Each particle, which represents a set of decision variables, is passed through this framework to construct and evaluate the airfoils and assign fitness. We used the baseline NLF0416 airfoil to obtain aspiration values, which are used to define the reference point. This reference point guides the swarm towards the preferred region of the objective landscape to find solutions of interest to the decision maker. The proficiency of the algorithm is highlighted by monitoring convergence and spread of solution using a hyper-volume calculation scheme suitable for user-preference based evolutionary many-objective algorithms. The results comparing the reference point based approach with a standard unguided non-dominated sorting based approach shows that the guided algorithm performs better in this many-objective problem instance. Final solutions found from the reference point based algorithm reveal an evident improvement over the NLF0416 airfoil across all operating conditions.