基于近似模型管理的微小卫星结构多目标优化设计  

Multi-objective Optimization Design for Microsatellite Structure Based on Approximation Model Management

作　　者：郑侃[1] 廖文和[1] 张翔[1] 

机构地区：[1]南京航空航天大学,南京210016

出　　处：《中国机械工程》2012年第6期655-659,共5页China Mechanical Engineering

摘　　要：针对某型号微小卫星结构多目标优化的特性,在基于信赖域近似模型管理的基础上,提出了一种新型的近似模型管理框架。选取正交试验法和Kriging近似法用于构建近似模型,选取基于拥挤距离的粒子群多目标优化算法作为核心算法,并通过最大最小距离策略来保证信赖度计算的准确性与合理性,从而确保获取与高精度模型一致的非劣解集。所选的五组优化结果显示,卫星结构质量均有不同程度的减小,最大减幅达到14.1%,同时整星前三阶固有频率较优化前明显提高。该近似模型管理框架在微小卫星结构多目标优化设计中的成功运用验证了其有效性及合理性,对于复杂工程多目标优化问题具有一定的工程应用价值。Aecording to the characteristics of multi--objective optimization for microsatellite structure, a new method for approximation model management was presented based on the trust region. Orthogonal and Kriging approximate method were used for the approximation model. The crowding distance in multiobjective particle swarm optimization was selected as the core algorithm. Meanwhile,the max--rain distance algorithm was used for ensuring the accuracy and rationality for trust measure- ment. Thus the results consistent with the real model could be obtained. The optimization results from five groups indicate that the structure mass is decreased at different degrees. The maximum amplitude reduction reaches 14.1 ~. Meanwhile,the first three orders of natural frequencies for the mierosatellite are improved obviously. The successful applications of the approximation model management frame in multi--objective optimization for satellite structure verify the effectiveness and the rationality, which can provide application values for complex engineering multi--objective optimization.

关 键 词：微小卫星 多目标优化设计 KRIGING模型 近似模型管理 

分 类 号：V423[航空宇航科学与技术—飞行器设计]