遗传算法及其在气动优化设计中的应用研究

【作者】 王晓鹏；

【导师】 高正红；

【作者基本信息】 西北工业大学 ， 飞行器设计， 2000， 博士

【摘要】 传统的确定性优化方法其本质决定了优化设计结果的局部优化特点，而遗传算法除了具有强的鲁棒性和并行性之外，还具有全局性优化的特点，这使得遗传算法在工程优化中得到越来越广泛的应用。然而，当前较精确的气动分析其计算量往往是十分庞大的，采用遗传算法作为优化方法会导致设计代价的大大增加，这对气动优化设计是十分不利的，甚至会降低遗传优化的可行性。为了克服遗传算法在气动优化设计中存在的弊端，提高优化设计的效率，本文着重对遗传算法进行研究和改进，建立了一些适用于气动优化设计的遗传优化模型，并将其应用于翼型、机翼和全机的气动优化设计。 本论文主要完成了以下几方面的工作： 1、针对标准遗传算法中存在的一些问题，对以二进制编码和解码技术为基础的标准遗传算法进行改进，采用实数编码技术，引入了非线性排名选择机制，应用优选技术、动态惩罚等技术，建立了适合气动优化设计的基本遗传优化模型。 2．把可变误差多面体方法与遗传算法结合起来，形成了松散式和紧密式两种不同性质的混合遗传算法，建立了气动优化设计中的混合遗传优化模型，提高了优化设计的效率和质量。 3．把Pareto方法与基于实数编码的遗传算法结合起来，形成了Pareto遗传算法，建立了用于处理多目标优化设计问题的方便高效的Pareto遗传优化模型。 4．通过求解二维和三维Euler方程，建立了翼型和机翼的气动分析模块；以工程分析方法为基础，建立了全机的气动分析模块。 5．建立了以Fortran Powerstation 4.0为开发环境的各种遗传优化模型的数值试验平台，对各遗传算法、改进技术及其优化模型的性能（优化质量和优化效率、可操作性）进行定量分析，并在此基础上建立了翼型、机翼和全机的优化设计软件系统。 6．应用建立的气动优化设计软件系统进行翼型、机翼和全机的气动优化设计，考察各种遗传优化模型对气动优化设计的适应性，检验优化前后气动性能的改善程度。 摘 要 通过对遗传算法进行的改进以及在气动优化设计中的试验性应用研究，得出的一些有用的结论，从而为借助数值优化方法进行飞机气动优化设计提供参考。更多还原

【Abstract】 Genetic algorithm is a global search algorithm based upon the mechanics of natural evolution. It shows its robustness in handling some complicated optimization problems. However, genetic algorithm requires a large number of evaluations of the objective function which normally involves executing a numerical solver of the governing equations. Therefore aerodynamic optimization design using genetic algorithm is computationally inefficient, even infeasible because of excessive computation cost. The purpose of the research here is to improve the performances of genetic algorithm especially when it is used in aerodynamic optimization design. The following work is performed: 1. The standard genetic algorithm is modified to form the genetic optimization model base on real number encoding. In the model established here, nonlinear ranking selection and dynamic penalty strategy are introduced. 2. Two types of hybrid genetic algorithms, loose hybrid genetic algorithm and tight hybrid genetic algorithm, are established by combining genetic algorithm with flexible tolerance polyhedron method. Due to their high computation efficiency, the hybrid genetic algorithms are suitable for dealing with aerodynamic optimization design with complex configuration. 3. Pareto genetic algorithm is formed to handle multi-objective optimization problems by combining genetic algorithm with pareto strategy. Compared with conventional approach, pareto genetic optimization are capable of dealing with multi-objective optimization problems more conveniently and more efficiently. 4. 2-D and 3-D Euler equations solvers are taken as the aerodynamic analysis tools for airfoil and wing. Axelson抯 engineering Abstract evaluation method is taken as the aerodynamic analysis tool for aircraft. 5. To analyze quantitatively the performances of genetic algorithms used in aerodynamic optimization design and the effects of the skills introduced, the numerical test planform for genetic optimization models is set up on Microsoft Fortran Powerstation. As applications, several aerodynamic optimization designs for airfoil, wing and aircraft are carried out to testify the adaptation of genetic optimization models. It can be concluded that the genetic algorithms developed in this paper have better performances when it is used in aerodynamic optimization designs of airfoil, wing and aircraft. Moreover, much higher efficiency will make genetic algorithms suitable for aerodynamic optimization design with more complex configuration.更多还原