基于支持向量机的卫星动力学多目标优化  

Multi-objective dynamic optimization of a satellite based on support vector machine

作　　者：袁野[1] 陈昌亚[2] 王德禹[1] 

机构地区：[1]上海交通大学海洋工程国家重点实验室,上海200240 [2]上海卫星工程研究所,上海200240

出　　处：《振动与冲击》2013年第22期189-192,共4页Journal of Vibration and Shock

摘　　要：针对实际工程中优化大规模结构耗时长问题,采用支持向量机与遗传算法相结合的优化方法。基于支持向量机建立结构近似模型,代替耗时长的有限元分析并应用遗传算法进行结构优化,形成省时、精度高、适应性强的优化方法。以某卫星整星结构动力学多目标优化为例,建立支持向量机近似模型模拟卫星蜂窝板、碳纤维框架结构设计参数与加速度频率响应、应力响应等非线性映射关系,选用二阶响应面模型作对比。计算结果表明,支持向量机与遗传算法组合优化方法不仅能降低耗时成本,且具精度良好。Aiming at the time-consuming problem of a large-scale structure＇s optimization, a method called SVM- GA combining support vector machine （SVM） and genetic algorithm （GA） was proposed, With this method, an approximate model was constructed based on SVM to replace the time-consuming FEA. GA was applied for optimization. A multi-objective structural dynamic optimization of a satellite was taken as an example to illustrate the effectiveness of the SVM-GA method. An approximate model based on SVM was buih to simulate the nonlinear mapping relations among structural design parameters of honeycomb plates and carbon fiber frames, and frequency responses of acceleration and stress of the satellite. The second order response surface model was adopted as a contrast. Results showed that tbe SVM- GA method can reduce the time consumption of optimization and has a good precision.

关 键 词：支持向量机 近似模型 遗传算法 优化设计 SUPPORT VECTOR MACHINE (SVM) GENETIC algorithm (GA) 

分 类 号：O24[理学—计算数学] O32[理学—数学]