Multi-modal distribution crossover method based on two crossing segments bounded by selected parents applied to multi-objective design optimization

다국어 초록 (Multilingual Abstract)

This paper discusses airfoil design optimization using a genetic algorithm (GA) with multi-modal distribution crossover (MMDX).

The proposed crossover method creates four segments from four parents, of which two segments are bounded by selected parents andtwo segments are bounded by one parent and another segment. After these segments are defined, four offsprings are generated. Thisstudy applied the proposed optimization to a real-world, multi-objective airfoil design problem using class-shape function transformationparameterization, which is an airfoil representation that uses polynomial function, to investigate the effectiveness of this algorithm. Theresults are compared with the results of the blend crossover (BLX) and unimodal normal distribution crossover (UNDX) algorithms. Theobjective of these airfoil design problems is to successfully find the optimal design. The outcome of using this algorithm is superior tothat of the BLX and UNDX crossover methods because the proposed method can maintain higher diversity than the BLX and UNDXmethods. This advantage is desirable for real-world problems.

참고문헌 (Reference)

1 B. M. Kulfan, "“Fundamental” parametric geometry representations for aircraft component shapes" 2006

2 M. Drela, "XFOIL 6.9 User Primer, Department of Aeronautics and Astronautics" Massachusetts Institute of Technology 2001

3 M. Drela, "Viscous-inviscid analysis of transonic and low Reynolds number airfoils" 25 (25): 1347-1355, 1988

4 K. Hijime, "Theoretical analysis of the unimodal normal distribution crossover for real-coded genetic algorithms" 2 (2): 187-194, 2002

5 L. J. Eshelman, "Real-coded genetic algorithms and interval-schemata, foundations of genetic algorithms 2" Morgan Kaufman Publisher 187-202, 1993

6 A. Ariyairt, "New multi-parent crossover based on crossing two segments bounded by selected parents" 126-137, 2014

7 T. Shigeyoshi, "Multi-parent recombination with simplex crossover in real coded genetic algorithms" 657-664, 1999

8 E. Zitzler, "Evolutionary algorithms for multiobjective optimization optimization: Method and application" Swiss Federal Institute of Technology (ETH), 1999

9 K. Kobayashi, "Discussing of clustering and network inversion for multi-objective genetic algorithms" 49 : 401-, 2007

10 K. Deb, "A fast and elitist multiobjective genetic algorithm:NSGA-II" 6 (6): 182-197, 2002