对混合遗传算法进行了初步研究,并以此为基础建立了压气机叶型优化设计平台.在遗传算法中引入模拟退火算法,增强了算法的局部寻优能力,提高了运行效率和优化质量.为维护群体的多样性,保证寻优的收敛,选择概率和交叉概率的设计可以随个体适应度和进化阶段的不同而自适应变化.二维叶型定义采用的是Bezier函数参数化定义方法.该方法可以较好的拟合叶型曲面,并通过少数控制点的调节灵活有效的修正叶型形面.对某压气机二维叶型的正问题数值优化结果表明本研究所建立的优化设计平台具有高效、可靠性好的特点.
A preliminary research is carried out in the genetic simulated annealing algorithm and a numerical optimization platform is developed based on the hybrid genetic algorithm. The introduction of simulated annealing into genetic algorithm notably improves the efficiency and capability of optimization. The adaptive crosser and mutation operators are devised to maintain the variety of population and to ensure the convergence of algo rithm. Having few control points, but allowing flexible manipulation, the Bezier representation is chosen to describe the two-dimensional compressor blade elements. The numerical optimization results show that the proposed numerical optimization platform works well with reasonable efficiency and robustness.