改进型卫星多学科设计优化算法  

Improved multidisciplinary design collaborative optimization for satellite design

作　　者：王梦赑[1] 张高飞 尤政[1] 

机构地区：[1]清华大学精密测试技术及仪器国家重点实验室,北京100084

出　　处：《清华大学学报（自然科学版）》2014年第11期1486-1491,共6页Journal of Tsinghua University(Science and Technology)

基　　金：国家"八六三"高技术项目(2013AA122601)

摘　　要：针对传统卫星多学科优化(MDO)方法存在优化问题容易陷入局部最优或无法收敛、计算量过大等缺陷,该文通过改善优化模式、引入遗传算法(GA)的方式对卫星协作优化方法(CO)进行了改进。针对协作优化算法在优化过程中容易陷入局部最优的问题,提出了一种基于局部搜索过程的混合遗传算法(GALS)。使用经典的Rosenbrock函数问题进行了性能测试,发现该算法是一种柔性灵活、能不断扩展和进步的开放式算法,能在保持各学科自治的基础上,异步并行地搜索系统最优解。该算法在实际卫星设计中已进行了多次应用验证,均取得了比传统优化算法更好的效果。Multidisciplinary design optimization （MDO） of overall satellite design and optimization can easily fall into a local optimum or not converge since the calculation is quite complicated. The tradi- tional optimization method is improved here by improving the optimi- zation model and using a genetic algorithm. A hybrid genetic algo- rithm based on a local search procedure prevents the optimization from falling into a local optimum. The Rosenbrock problem used to test the performance of the improved collaborative optimization vali- dates the improved computing properties. The improved collabora- tive optimization method is a flexible open method which can be con- tinuously expanded and developed. The system can search for the optimal solution asynchronously in parallel to preserve the autonomy of the various disciplines. This algorithm has been applied many times to satellite design with good results, better than traditional op- timization algorithms.

关 键 词：卫星总体设计 多学科设计优化(MDO) 协作优化方法(CO) 遗传算法(GA) 局部搜索遗传算法(GALS) 

分 类 号：V423.4[航空宇航科学与技术—飞行器设计]