基于Agent模型的多学科设计优化方法

复杂工程系统的设计往往涉及多门学科 ,多学科设计优化 (MDO)方法是一类求解复杂系统设计优化问题的方法。本文首先简要回顾现有 MDO方法的特点 ,然后基于 Agent模型提出了一种新的多学科设计优化方法——子空间自主式优化方法。这种方法具有如下特点 :流程简单 ;各个学科组可自主地、并行地进行设计和优化 ;灵活性强 ;所需系统分析的计算次数少 ;没有系统级协调优化环节。为了验证这种算法的有效性 ,对二个典型算例进行了数值实验。初步的数值实验结果表明 ,二个典型的算例经过几次迭代后均能收敛于最优解。子空间自主式方法是一种较有潜力的 MDO方法 ,值得进一步研究

A Method of Multidisciplinary Design Optimization Using Agent-Based Model

The design of highly complex engineering systems often involves several different disciplines. The methods of multidisciplinary design optimization (MDO) provide the framework for solving design optimization of complex engineering systems. The current MDO methods are first reviewed briefly. A novel framework of multidisciplinary design optimization is proposed using agent based model, which is called the subspace autonomous optimization. This new MDO method has several advantages: (1) simplicity in organization, (2) disciplinary autonomy and concurrency in design optimization, (3) flexibility for a variety of disciplinary design optimization, and (4) considerable reduction of computational burden for system analysis. Two numerical test problems are used to verify the method. The preliminary results of this method are quite encouraging.