Multi-Objective Optimization With Adaptive Genetic Algorithms

Non-dominated sorting genetic algorithm(NSGA-2) is one of the classic algorithm for solving multi-objective optimization problems. Crowding distance adopted in NSGA-2 plays an critical role in both convergence and diversity. In order to improve the uniformity of solution sets so as to shorten the distance between solution sets and Pareto optimal front, some details are modified about crowding distance and the concept is introduced about “bilateral range difference”,based on adaptive crossover and mutate operator, added with minimum distance filtration, an improved adaptive NSGA-2 algorithm is suggested. Secondly, the proposed algorithm is verified by experiment on four multi-objective optimization test functions on the aspects of convergence and diversity. Finally, the developed IANSGA-2 is applied for optimizing a classic airfoil with the target of maxmuning both the lift coefficient and lift-drag ratio with different speed and the airfoil shape is deviced in different status. The result indicates that the lift coefficient and the drag coefficient are distributed uniformly on Pareto front after optimization. In addition, comparing with the initial airfoil, the aerodynamic performance has been improved obviously. Consequently, the present developed adaptive multi-objective optimization algorithm is better than classic NSGA-2.