基于DAKOTA的多学科优化计算平台的构建与应用

发布时间：2018-05-14 16:33

本文选题：DAKOTA + 多学科优化框架　；参考：《上海交通大学》2014年硕士论文

【摘要】：多学科设计优化是综合多种学科知识，进行优化设计，获取综合多学科特点的可行优化结果的过程。而多学科优化平台，就是指构建一种能实现多学科设计优化的方法的计算环境，它能集成运行不同学科的计算，实现设计、分析与优化过程的集成与自动化。随着工程问题的复杂化，企业和研究团体都迫切希望能够通过多学科优化平台的辅助，缩短优化时间，提高效率和产品的技术含量，以适应激烈的市场竞争。在此背景下，本文以开源的大型工程优化框架DAKOTA为算法库，在Orange下构建一个支持可视化编程、优化问题分析、处理、实时优化监控、数据挖掘等功能的多学科优化平台。提出并开发了多学科优化平台的6个组件，分别是master(总控单元)、model generator(模型生成单元)、optimizer(优化器单元)，monitor(监视器单元)、data manager（数据管理单元）、post processor(后处理单元)。着重研究了Pro/E、NX、ANSYS等平台下的自动参数化技术，以及相关软件的接口开发技术，实现了在多学科优化平台中集成CAD、CAE组件的目标。通过燃气轮机轮盘优化和水辅成型两个多学科优化实例对本平台实际效果进行了检验。经检测，这一平台能够驱使平台组件进行迭代寻优，使整个多学科优化过程更为自动化、智能化、便捷化。此平台在效率、开源性、可扩展、费用等方面，，拥有相当的竞争力。同时由于能够实现设计、分析与优化过程的自动化，降低对工程技术人员多学科背景知识的依赖，有利相关技术在企业的推广应用，具有重要的实际应用价值。
[Abstract]:Multidisciplinary design optimization is the process of synthesizing the knowledge of many disciplines, carrying on the optimization design, and obtaining the feasible optimization result of synthesizing the multi-disciplinary characteristics. The multi-disciplinary optimization platform is to construct a computing environment that can realize the method of multidisciplinary design optimization. It can integrate and automate the process of design, analysis and optimization by integrating and running the calculation of different disciplines. With the complexity of engineering problems, enterprises and research groups are eager to shorten the optimization time, improve the efficiency and the technical content of products through the assistance of multi-disciplinary optimization platform, so as to adapt to the fierce market competition. In this context, the open source large-scale engineering optimization framework (DAKOTA) is used as the algorithm library, and a multidisciplinary optimization platform is constructed under Orange to support the functions of visual programming, optimization problem analysis, processing, real-time optimization monitoring, data mining and so on. Six components of multidisciplinary optimization platform are proposed and developed, which are master (master) model generator (monitor unit) and monitor (data management unit). The automatic parameterization technology based on Pro-EX NX ANSYS and the interface development technology of related software are studied emphatically. The goal of integrating CADU CAE components in multidisciplinary optimization platform is realized. The practical effect of this platform is tested by two multidisciplinary optimization examples of gas turbine disc optimization and water assisted molding. After testing, the platform can drive the platform components to iterative optimization, making the whole multidisciplinary optimization process more automatic, intelligent and convenient. This platform is competitive in efficiency, open source, extensibility, cost and so on. At the same time, it can realize the automation of design, analysis and optimization process, reduce the dependence on engineering and technical personnel's multi-disciplinary background knowledge, and benefit the popularization and application of related technology in enterprises, which has important practical application value.
【学位授予单位】：上海交通大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TB47;TK472

【参考文献】