作战仿真混合实验方案设计研究

doi:10.16182/j.issn1004731x.joss.22-1309

摘要/Abstract

摘要：

作战仿真实验方案设计指的是基于基准的作战想定，利用各种实验设计方法对实验因子的取值进行抽样，形成一组实验方案，供后续仿真使用。作战仿真的复杂性包括实验因子数目众多，因子类型多样，既有连续数值类型，又有离散数值类型，这对高效的混合实验设计方法提出了迫切的需求。针对这些需求，对作战仿真混合实验方案设计技术展开了研究。对实验设计方法做了简单的分类和综述，给出了3种混合实验方案设计方法的流程及其优缺点分析；介绍了一个面向作战仿真的实验设计系统，包括软件架构设计、功能组成及实验设计流程；以一个作战想定为例，演示和分析了3种混合实验方案设计方法的具体应用。上述方法和工具将有效指导复杂作战仿真中具有离散和混合类型因子的实验设计问题，并在多个仿真项目得到了应用和验证。

关键词: 作战仿真, 想定, 实验方案设计, 混合实验设计, 实验设计系统

Abstract:

Combat simulation experimental design refers to sampling the values of experimental factors based on baseline combat scenarios using various experimental design methods and then generating a set of experimental schemes for the sequential simulation. The complexity of combat simulation, such as numerous experimental factors and distinct factor types, including continuous and discrete numeric types, poses several challenges and requires efficient hybrid experimental design methods. To address these issues, this paper conducts a study on the hybrid experimental scheme design for combat simulation. This paper gives a brief classification and review of experimental design methods and presents three hybrid experimental scheme design methods with their merits and drawbacks analyzed. The paper introduces an experimental design system for combat simulation, including its software architectural design, functions, and experimental design procedure and demonstrates and analyzes the application of three hybrid experimental scheme design methods in a combat scenario. The tool and methods presented in this paper will effectively guide the experimental design problem in complex combat simulation with discrete and mixed type factors, and it has been applied and verified in several simulation projects.

Key words: combat simulation, scenario, experimental scheme design, hybrid experimental design, experimental design system

中图分类号:

TP391.9

刘飞,赖鹏,陆营波等 . 作战仿真混合实验方案设计研究[J]. 系统仿真学报, 2024, 36(3): 735-742.

Liu Fei,Lai Peng,Lu Yingbo,et al . Research on Hybrid Experimental Scheme Design for Combat Simulation[J]. Journal of System Simulation, 2024, 36(3): 735-742.

图/表 11

图1

图2

图3

表1

图4

图5

图6

表2

表3

表4

表5

参考文献 17

1	孔晨妍, 崔鹏. 联合作战仿真实验设计方法研究[C]//第八届中国指挥控制大会论文集. 北京: 兵器工业出版社, 2020: 674-679.
2	杨雪生, 刘云杰, 李梦汶. 联合作战仿真实验的设计与开发[J]. 系统仿真学报, 2011, 23(7): 1522-1526.
	Yang Xuesheng, Liu Yunjie, Li Mengwen. Design and Development of Simulation Experimentation for Joint Warfare[J]. Journal of System Simulation, 2011, 23(7): 1522-1526.
3	赵选民. 试验设计方法[M]. 北京: 科学出版社, 2006.
4	Wang Shuo, Liye Lü, Du Li, et al. An Improved LHS Approach for Constrained Design Space Based on Successive Local Enumeration Algorithm[C]//2019 IEEE 9th Annual International Conference on CYBER Technology in Automation, Control, and Intelligent Systems (CYBER). Piscataway, NJ, USA: IEEE, 2019: 896-899.
5	Mahto Navin, Chakravarthy Satyanarayanan R. Response Surface Methodology for Design of Gas Turbine Combustor[J]. Applied Thermal Engineering, 2022, 211: 118449.
6	Chen R B, Hsieh D N, Hung Y, et al. Optimizing Latin Hypercube Designs by Particle Swarm[J]. Statistics and Computing, 2013, 23(5): 663-676.
7	耿国庆, 王成皓, 段利斌, 等. 一种改进的约束域拉丁方抽样算法及工程应用[J]. 重庆理工大学学报(自然科学), 2020, 34(10): 58-66.
	Geng Guoqing, Wang Chenghao, Duan Libin, et al. Formulation and Engineering Application of an Improved Constrained Domain Latin Hypercube Sampling Algorithm[J]. Journal of Chongqing University of Technology(Natural Science), 2020, 34(10): 58-66.
8	Shang Xiaobing, Chao Tao, Ma Ping, et al. An Efficient Local Search-based Genetic Algorithm for Constructing Optimal Latin Hypercube Design[J]. Engineering Optimization, 2020, 52(2): 271-287.
9	Wang Lin, Sun Fasheng, Lin D K J, et al. Construction of Orthogonal Symmetric Latin Hypercube Designs[J]. Statistica Sinica, 2018, 28(3): 1503-1520.
10	Su Zheren, Wang Yaping, Zhou Yingchun. On Maximin Distance and Nearly Orthogonal Latin Hypercube Designs[J]. Statistics & Probability Letters, 2020, 166: 108878.
11	Qian P Z G. Sliced Latin Hypercube Designs[J]. Journal of the American Statistical Association, 2012, 107(497): 393-399.
12	Chen Hao, Yang Jinyu, Lin D K J, et al. Sliced Latin Hypercube Designs with Both Branching and Nested Factors[J]. Statistics & Probability Letters, 2019, 146: 124-131.
13	Wang Xiaodi, Chen Xueping, Lin D K J. Sliced Symmetrical Latin Hypercube Designs[J]. Journal of Statistical Planning and Inference, 2022, 218: 59-72.
14	Ba Shan, Myers W R, Brenneman W A. Optimal Sliced Latin Hypercube Designs[J]. Technometrics, 2015, 57(4): 479-487.
15	Joseph V R, Gul E, Ba Shan. Designing Computer Experiments with Multiple Types of Factors: the MaxPro Approach[J]. Journal of Quality Technology, 2020, 52(4): 343-354.
16	Kang Xiaoning, Deng Xinwei. Design and Analysis of Computer Experiments with Quantitative and Qualitative Inputs: a Selective Review[J]. WIREs Data Mining and Knowledge Discovery, 2020, 10(3): e1358.
17	祝颂, 钱晓超, 陆营波, 等. 基于XGBoost的装备体系效能预测方法[J]. 空天防御, 2021, 4(2): 1-6.
	Zhu Song, Qian Xiaochao, Lu Yingbo, et al. An XGBoost-based Effectiveness Prediction Method of Equipment System-of-systems[J]. Air & Space Defense, 2021, 4(2): 1-6.

方法	优点	缺点
基于多种实验设计方法的混合分组因子抽样实验	分组灵活，可将因子分为任意离散和连续子集，每个子集采用任何一种实验设计方法	产生实验方案数目过多，需要运行大量仿真实验；样本空间原则上不满足整体均匀性
基于分片拉丁超立方设计的混合因子抽样实验	为每种离散类型因子的水平组合设计一个不同的拉丁超立方分片；充分考虑每个分片和整体的空间填充均匀性	只能处理较少的离散类型因子；分片数大时实验次数过多，产生实验方案数目过多
基于最大投影实验设计的混合因子抽样实验	可产生较少的实验方案数目；能处理更多的离散类型因子；设计结果有良好的整体均匀性和二维投影均匀性	方法本身较复杂，但通过封装在工具中，可以克服

编号	名称	因子类型	因子水平数	水平设置
X₁	单目标发射导弹数量	离散	2	1和2
X₂	使用火力通道数	离散	2	4和8
X₃	雷达发现概率	连续	4	均匀分布[0.7,0.9]
X₄	战斗准备时间	连续	4	均匀分布[1,8]
X₅	单发导弹发射时间间隔	连续	4	均匀分布[1,4]
X₆	拦截斜距	连续	4	均匀分布[6 000,25 000]
X₇	发射导弹时间间隔	连续	4	均匀分布[1,4]
X₈	目标RCS	连续	4	均匀分布[0.1,1.9]

序号	X₁	X₂	X₃	X₄	X₅	X₆	X₇	X₈
1	1	4	0.82	7.63	1.03	19 243.90	2.95	1.18
2	1	4	0.71	3.16	3.20	23 245.30	3.29	0.95
3	1	4	0.80	1.07	2.06	9 980.30	1.68	1.68
4	1	4	0.90	4.99	3.72	13 361.40	2.40	0.36
5	2	4	0.85	6.43	2.59	21 724.10	1.36	1.28
6	2	4	0.79	4.18	2.27	12 964.40	3.75	1.88
7	2	4	0.75	4.77	3.27	6 653.18	2.01	0.56
8	2	4	0.85	2.33	1.25	18 859.50	2.81	0.24
9	1	8	0.77	6.98	3.99	15 934.30	2.51	1.60
10	1	8	0.80	1.88	2.95	11 445.40	3.91	0.48
11	1	8	0.73	3.77	1.88	20 513.20	1.38	0.70
12	1	8	0.89	5.54	1.62	7 220.82	2.15	1.06
13	2	8	0.84	1.68	3.62	14 975.50	1.00	0.85
14	2	8	0.72	5.95	1.40	8 472.10	1.90	1.50
15	2	8	0.75	7.34	2.70	16 781.10	3.23	0.18
16	2	8	0.87	3.26	2.38	24 000.90	3.58	1.41

序号	X₁	X₂	X₃	X₄	X₅	X₆	X₇	X₈
1	1	4	0.81	4.29	3.78	9 735.99	1.22	1.49
2	1	4	0.80	1.96	2.83	24 577.30	3.91	1.00
3	1	4	0.71	5.31	1.25	11 319.60	2.98	1.29
4	1	4	0.88	6.67	1.91	17 812.00	2.29	0.29
5	2	4	0.77	6.09	3.32	9 283.48	1.59	0.20
6	2	4	0.83	6.73	2.48	22 826.20	2.08	1.58
7	2	4	0.71	1.00	3.10	16 319.30	2.63	1.39
8	2	4	0.89	3.28	1.68	12 413.50	3.70	0.82
9	1	8	0.74	7.37	2.91	21 340.30	3.17	0.61
10	1	8	0.85	3.67	3.46	13 680.00	3.45	1.71
11	1	8	0.78	1.76	2.06	19 881.10	1.51	0.46
12	1	8	0.84	5.54	1.48	7 050.16	1.82	1.12
13	2	8	0.86	4.57	3.91	21 564.20	2.44	0.75
14	2	8	0.81	3.15	1.18	18 788.40	2.74	1.88
15	2	8	0.73	7.82	2.24	14 725.40	1.15	1.05
16	2	8	0.76	2.42	2.66	7 396.35	3.30	0.37

序号	X₁	X₂	X₃	X₄	X₅	X₆	X₇	X₈
1	1	4	0.90	5.02	2.96	19 565.80	3.76	1.07
2	2	4	0.76	1.82	1.69	17 919.90	2.00	0.15
3	1	8	0.80	6.60	1.50	21 431.90	2.20	1.64
4	2	8	0.73	5.55	2.45	22 360.10	3.18	0.39