基于可能性理论的认知不确定性条件下多学科设计优化研究

项目名称： 基于可能性理论的认知不确定性条件下多学科设计优化研究

项目编号： No.50805018

项目类型： 青年科学基金项目

立项/批准年度： 2009

项目学科： 建筑科学

项目作者： 杨波

作者单位： 电子科技大学

项目金额： 20万元

中文摘要： 本项目对认知不确定性条件下的多学科设计优化(MDO)开展了深入的研究。提出了一种基于最大确定性原则的可能性分布(PD)的构造方法(MSBPD)，减小了PD构造过程中信息的损失。提出了一种小样本条件下的PD构造方法。提出了一种新的评判PD构造方法在最大确定性程度上表现的方法。提出了一种同时包含两种(随机和认知)不确定性的设计优化模型及求解算法(SOCUA算法)，降低了设计优化过程的计算量。提出了一种认知不确定条件下的MDO模型(PBMDO-SOPA)以及三种基于可能性的MDO方法(PBMDO-SOPA-MDF、PBMDO-SOPA-SAND、PBMDO-SOPA-IDF)。提出了随机/认知/两种不确定性条件下的稳健可靠性多目标设计优化模型，并提出了一种基于顺序优化和可靠性/可能性估计(SORPA)和物理规划的模型求解算法，能同时兼顾设计的可靠性和稳健性。提出了一种同时包含两种不确定性的基于协同优化和SORPA的MDO模型和优化算法。提出了一种基于模糊神经网络的智能交互式物理规划方法。提出了一种基于智能交互式物理规划的MDO方法。本项目的研究成果对MDO理论的发展与工程应用具有重要的意义。

中文关键词： 认知不确定性；可能性分布；多学科设计优化；顺序优化和可靠性/可能性估计；协同优化

英文摘要： In this project, we've conducted in-depth research on multidisciplinary design optimization (MDO) under epistemic uncertainty (EU). A new method of deriving possibility distribution (PD) based on maximal specificity principle was proposed, which can reduce the loss of information in PD derivation. A new method of deriving PD under small sample size was proposed. A new method of assessing the degree of maximal specificity of PD derivation methods was proposed. A new design optimization (DO) model under both aleatory uncertainty (AU) and EU as well as the corresponding algorithm was proposed, which requires less computational effort. An MDO model under EU was proposed, and three types of corresponding algorithm which are based on possibility theory were established. Models of robust-reliability multi-objective DO under AU/EU/both AU and EU are proposed, and the corresponding algorithm based on sequential optimization and reliability/possibility assessment (SORPA) and physical programming (PP) was established, which can take into count both reliability and robust factors. An MDO model under both AU and EU was proposed, and the corresponding algorithm was established. The model and algorithm are based on collaborative optimization (CO) and SORPA. An intelligent interactive physical programming (IIPP) method based on fuzzy neural networks was proposed, and an MDO model based on IIPP was proposed. The research findings of this project have important impact on MDO theory as well as its industrial application.

英文关键词： Epistemic uncertainty; Possibility distribution; MDO; Sequential Optimization and Reliability/Possibility Assessment; Collaborative Optimization