项目名称: 基于本征正交分解的多学科设计优化场量耦合求解策略研究
项目编号: No.51505385
项目类型: 青年科学基金项目
立项/批准年度: 2016
项目学科: 机械、仪表工业
项目作者: 粟华
作者单位: 西北工业大学
项目金额: 20万元
中文摘要: 随着基于数值计算的高拟真度分析方法在多学科设计优化中的广泛使用,场量作为其典型特征出现在越来越多的学科交叉耦合当中。传统的整体式和分布式多学科设计优化求解策略在面向包含场量的多学科设计优化问题时计算效率低下且求解困难,无法满足现代复杂系统的设计需求。本项目针对同时耦合场量和常规变量的多学科设计优化问题,采用非线性本征正交分解方法进行耦合场量降维,以实现分布式求解策略对场量的支持。通过分析场量多学科设计优化问题的耦合作用机理,推导不同耦合情况下的多学科设计优化数学模型,结合分布式求解架构提出主动/被动方式的场量增点策略,建立具有自适应能力的本征正交分解方法,并在此基础上发展基于自适应本征正交分解模型的协同优化与解析目标分流分布式求解策略,以达到降低场量多学科设计优化问题求解难度,提高计算效率的目的。本项目为复杂系统的多学科设计优化提供基础理论和方法支持。
中文关键词: 多学科设计优化;本征正交分解;场量耦合;求解策略;分布式架构
英文摘要: As the expanded use of the high-fidelity analysis model based on numerical calculation in Multidisciplinary Design Optimization (MDO), field variable appears on the interdisciplinary coupling more and more as the typical characteristic of the high-fidelity analysis models. Traditional MDO solution methods on monolithic architecture and distributed architecture have difficult convergence and low computation efficiency when facing the new MDO problem coupling with field variable, which cannot meet the high demands of the modern complex system design. In this project, the nonlinear Proper Orthogonal Decomposition (POD) method is used to reduce the dimension of the coupling field variable to give a better support of the distributed MDO solution method for the MDO problem coupling with field variables and normal variables simultaneously. The MDO mathematical model in different coupling conditions are derived from the coupling relationship analysis of the MDO problem with field variables, the initiative and passive adding points strategy is proposed based on the distributed architecture, a modified POD method with adaptive update capability is created, and on this basis the Collaborative Optimization (CO) and Analytical Target Cascading (ATC) distributed MDO solution methods with adaptive POD model is developed to reduce the solving difficulty and increase the computation efficiency. This project has some motivating foundation support on MDO method in complex systems.
英文关键词: Multidisciplinary design optimization;Proper orthogonal decomposition;Field variable coupling;Solution method;Distributed architecture