Improved intelligent genetic algorithm applied to long-endurance airfoil optimization design

Abstract

This study presents a genetic algorithm (GA) with simulated annealing (SA) mutation that has been improved by including a fractional factorial design for the crossover operator and SA for the mutation operator. The proposed GA generates the chromosomes of children using an intelligent crossover process with factorial experiments, rather than exchanging the genes at random using the one-point, two-point, multi-point, or uniform crossover employed in a traditional GA. Therefore, the GA with the intelligent crossover operator is referred to as the intelligent genetic algorithm (IGA) in this article. The SA mutation was employed to replace the conventional jump mutation in order to enhance the search process and avoid individuals becoming trapped in local optima. The performance of the proposed IGA was assessed via a nonlinear multimodal function with two design variables. Four representative test cases (Sphere, Rosenbrock, Rastrigin and Griewank functions) by used the IGA with SA mutation, micro-IGA, and IGA with jump mutation to evaluate the capacity and efficiency of the proposed IGA in terms of large design variables. Computational results of the convergence history and optimal solution for the benchmarking cases indicate that the proposed IGA with SA mutation outperforms the micro-IGA and IGA with jump mutation. Two profile fittings of NACA 4415 and LS(1)-0417 Modified airfoils are also performed using the proposed IGA with SA mutation. The proposed IGA is also applied to long-endurance airfoil optimization design.