系统工程与电子技术 ›› 2023, Vol. 45 ›› Issue (7): 2078-2085.doi: 10.12305/j.issn.1001-506X.2023.07.18

• 系统工程 • 上一篇    下一篇

基于高斯过程模型的定性定量因子混合补充试验设计方法

张路路, 潘正强, 刘天宇, 金光   

  1. 国防科技大学系统工程学院, 湖南 长沙 410073
  • 收稿日期:2022-01-24 出版日期:2023-06-30 发布日期:2023-07-11
  • 通讯作者: 潘正强
  • 作者简介:张路路(1988—), 男, 博士研究生, 主要研究方向为系统试验与评估
    潘正强(1981—), 男, 副教授, 博士, 主要研究方向为武器装备试验鉴定
    刘天宇(1989—), 男, 讲师, 博士, 主要研究方向为系统可靠性评估、装备试验鉴定
    金光(1973—), 男, 研究员, 博士, 主要研究方向为寿命预测与健康管理、系统试验与评估、数据分析与建模
  • 基金资助:
    国家自然科学基金(72171231);国家自然科学基金(72001210)

Supplemental experimental design method of qualitative-quantitative hybrid factor based on Gaussian process model

Lulu ZHANG, Zhengqiang PAN, Tianyu, LIU, Guang JIN   

  1. College of Systems Engineering, National University of Defense Technology, Changsha 410073, China
  • Received:2022-01-24 Online:2023-06-30 Published:2023-07-11
  • Contact: Zhengqiang PAN

摘要:

针对舰空导弹试验设计中同时存在定性、定量因子的问题, 提出了一种定性-定量因子混合的补充试验设计方法。首先, 构建定性-定量因子混合评估模型, 针对模型中定性、定量因子相关函数难以融合的问题, 利用超球面分解方法对定量因子相关函数进行定量化处理。然后, 在预测响应模型的基础上构建预测均方差探索项, 通过进一步计算样本点的预测误差值, 对探索项进行修正, 提高补充试验设计的开发能力。通过计算交叉多面体比率得到最佳邻域, 再利用遗传算法在最佳邻域中确定补充试验点。最后, 通过数值算例和示例分析, 验证了所提方法的有效性。

关键词: 定性-定量因子混合, 补充试验设计, 探索-开发, 最佳邻域

Abstract:

To solve the problem that qualitative and quantitative factors exist simultaneously in naval air defense missile experimental design, a supplementary experiment design method based on the hybrid of qualitative and quantitative factors is proposed. Firstly, a qualitative-quantitative factor fusion evaluation model is constructed, and in order to solve the problem that it is hard for the qualitative and quantitative factor correlation functions to fuse in the model above, a hypersphere decomposition method is used to quantify the quantitative factor correlation function. Then, based on the prediction response model, the prediction variance exploration items are constructed. By funther calculating the prediction error of sample points, the exploration item is corrected to improve the development ability of supplementary experiment design. The best neighborhood is obtained by calculating the cross polyhedron ratio, and then the supplementary experiment points are determined in the best neighborhood by the genetic algorithm. Finally, the effectiveness of the proposed method is verified by numerical examples and case analysis.

Key words: qualitative-quantitative hybrid factor, supplemental experiment design, exploration-exploitation, best neighborhood

中图分类号: