摘要
鉴于多学科系统不确定性分析常规方法——灵敏度分析方法的大信息量要求和计算过程的约束性,基于Bayes理论,利用验前信息计算子系统输出的验后分布,使用Gibbs抽样方法获得子系统输出的不确定性统计信息,进而利用不确定性信息改进协同优化算法(CO)方法的子系统优化,得到基于不确定性的CO方法。数值算例验证了Bayes方法在有限信息量下计算准确,效率高,适用范围广;导弹设计示例说明了基于不确定性的CO方法的有效性,保障了设计结果的可用性。
As for the request of large information and the restriction of calculation brought by sensitivity analysis method,which is normally used to analyze uncertainty in multidisciplinary system,Bayesian theory is introduced.In Bayesian method,prior information is used to achieve posterior distributions of subsystem output and Gibbs sampling method is used to calculate uncertainty statistical information of subsystem output.Then,modified Collaborative Optimization（CO） algorithm is educed by improving subsystem optimization with uncertainty.The numerical example is used to validate that Bayesian method is precise,efficient and widely applicable under limited information.An example of missile design proves that the CO algorithm based uncertainty is effective,which ensures the usability of design results.
出处
《导弹与航天运载技术》
北大核心
2010年第5期46-50,共5页
Missiles and Space Vehicles
