基于SAGE的大型客机DMDO系统的关键技术研究  

Study of key technologies in civil airplane DMDO system based on SAGE

作　　者：李莹[1] 李旭[1] 林新华[1] 刘洪[2] 李明禄[1] 陆鑫达[1] 

机构地区：[1]上海交通大学计算机系,上海200240 [2]上海交通大学航空航天学院,上海200240

出　　处：《计算机工程与设计》2010年第19期4288-4292,共5页Computer Engineering and Design

基　　金：国家自然科学基金项目(90205006);国家863高技术研究发展计划基金项目(2006AA01Z247)

摘　　要：为解决大型客机的设计过程中多方设计研究人员的协同工作问题,以及在不断调整设计参数的迭代设计过程中出现的各种问题,基于SAGE(simple adaptive grid engine)[1]的分布式多学科优化设计(distributed multidisciplinary design optimization,DMDO)系统中的3项创新及关键技术进行了深入的探讨:①MDO(multidisciplinary design optimization)[2]领域专用的工作流引擎,它提供灵活的封装模块和高效的封装方式;②异构集群的元作业调度器,根据MDO的应用特征进行简化,使用短作业优先的调度策略减少了作业的平均等待时间;③基于SAGE的分布式集群计算。针对这3项关键技术,在实验环境中分别进行了成本,等待时间和可扩展性的测试。实验结果表明,这3项关键技术都满足了预期的设计目标,能够为大型客机的专业设计人员提供一个高效实用的协同设计平台。To solve problems raised in the iteration procedure for tuning design parameters and collaboration among designers from multiple institutes during the design of the civil airplane, the distributed multidisciplinary design optimization （DMDO） system based on simple adaptive grid engine （SAGE） is designed and implemented. Three key technologies in DMDO system are covered： ① multidisciplinary design optimization （MDO） domain specific workflow engine, which provides flexible packaging modules and high efficiency packaging methods. ② meta-job scheduler for heterogeneous cluster, which reduces average job waiting time by using shnrtest job first policy, simplifies according to characters of MDO applications. ③Distributed cluster computing based SAGE. Cost, job waiting time and scalability are tested in experiments, and the results show that the three key technologies meet the expected design goals, which indicates DMDO system could provide the designers of civil airplane an efficient and flexible collaborative design platform.

关 键 词：大型客机 多学科优化 工作流 分布式计算 高性能计算 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]