Research Initiation: Global Optimization Algorithms for Engineering Design

研究启动：工程设计全局优化算法

• 批准号: 9211001
• 负责人: Zelda Zabinsky
• 金额: $ 9.63万
• 依托单位: University of Washington
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 1992
• 资助国家: 美国
• 起止时间: 1992-07-15 至 1996-06-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=9211001&HistoricalAwards=false
• 关键词: Research; Initiation; Global; Optimization; Algorithms

The engineering community needs efficient algorithms for solving global optimization problems in design. Global optimization problems require minimizing a function possessing multiple local minima. A practical example is finding the optimal design of a composite laminated stiffened panel used in aircraft structures. This research has three objectives. The first is to develop a random search algorithm that includes both integer and continuous variables. A simulated annealing algorithm for mixed integer/continuous global optimization will be developed, because simulated annealing has already demonstrated some effectiveness in both the integer domain and the continuous domain. The second is to take advantage of estimates of a Lipschitz constant in the optimization algorithm. The third is to decrease an algorithm's probability of getting trapped in a local minimum that is not the global minimum. Together these objectives will improve algorithms in a manner that will make formulating and solving global optimization problems more practical and accessible to engineers. The results will be applicable to aspects of design in all engineering disciplines.

工程界需要有效的算法来解决设计中的全局优化问题。全局优化问题要求最小化具有多个局部极小值的函数。以飞机结构中复合材料层合加筋板的优化设计为例。这项研究有三个目的。首先是开发一种随机搜索算法，该算法同时包含整数变量和连续变量。由于模拟退火算法已经在整数域和连续域证明了一定的有效性，因此将开发一种用于整数/连续混合全局优化的模拟退火算法。二是利用优化算法中Lipschitz常数的估计。第三是降低算法陷入局部最小值而非全局最小值的概率。总之，这些目标将以一种方式改进算法，使工程师能够更实际、更容易地制定和解决全局优化问题。结果将适用于所有工程学科的设计方面。