基于深度置信网络效能拟合的火控系统精度全局敏感性分析

doi:10.12382/bgxb.2023.0659

摘要/Abstract

摘要：

针对当前航空火控系统的精度研究对数据完备性要求较高,且仅能分析系统内单误差源影响因素的问题,提出一种全新的基于深度置信网络(Deep Belief Network,DBN)效能拟合的火控系统精度全局敏感性分析(Global Sensitivity Analysis via Deep Belief Network,GSADBN)算法。从全局敏感性分析算法的优劣点出发,分析传统的全局敏感性分析算法在不完备数据下存在的局限性。利用DBN优异的特征提取能力,并采用无监督训练和有监督微调相结合的算法,搭建并训练火控系统效能拟合模型。研究结果表明:与传统Sobol算法、经典傅里叶振幅敏感性检验(Fourier Amplitude Sensitivity Test, FAST)算法以及最新的基于贝叶斯网络的Sobol(Bayesian Neural Sobol, BNSobol)算法相比, GSADBN算法不仅可以满足精度要求,同时还可以在不完备数据下达到传统算法在完备数据下精度分析的效果;该算法可以在火控系统不完备数据下取得较好的精度分析结果,同时在设计火控系统各模块时,给出精度方面的合理方案,从而为火控系统的设计及作战效能的提高提供参考和理论支撑。

关键词: 航空火控系统, 深度置信网络, GSADBN算法, 精度评估, 全局敏感性分析

Abstract:

In view of the fact that most of the current researches on the accuracy of fire control system require high data integrity and only analyze the influence factors of single error source, this paper proposes a novel global sensitivity analysis (GSA) method based on deep belief network (DBN) effectiveness fitting for the accuracy of fire control system. Starting from the advantages and disadvantages of the traditional GSA methods, the limitations of the traditional methods in the case of incomplete data are analyzed. Then, a fire control system performance fitting model is constructed and trained by utilizing the excellent feature extraction ability of DBN and a combination of unsupervised training and supervised fine-tuning methods. The proposed GSADBN method is compared with traditional Sobol method, classical Fourier amplitude sensitivity test (FAST) method and improved BNSobol method. The experimental results show that the proposed GSADBN method can not only meet the accuracy requirements, but also obtain the excellent accuracy analysis results when the data from aviation fire control system is incomplete. Furthermore, GSADBN method can give a reasonable scheme of accuracy value when designing each module of fire control system, so as to provide reference and theoretical support for the design of fire control system and the improvement of combat effectiveness.

Key words: aviation fire control system, deep belief network, GSADBN method, accuracy assessment, global sensitivity analysis

中图分类号:

V19
V233.7

汪强龙, 高晓光, 李新宇, 闫栩辰, 万开方. 基于深度置信网络效能拟合的火控系统精度全局敏感性分析[J]. 兵工学报, 2024, 45(10): 3430-3444.

WANG Qianglong, GAO Xiaoguang, LI Xinyu, YAN Xuchen, WAN Kaifang. Global Sensitivity Analysis on Accuracy of Aviation Fire Control System via DBN Effectiveness Fitting[J]. Acta Armamentarii, 2024, 45(10): 3430-3444.

图/表 16

参考文献 28

[1]	陆彦. 航空火力控制技术[M]. 西安: 西北工业大学出版社, 1989: 24-50, 221-226, 296-304.
	LU Y. Aviation fire control technology[M]. Xi’an: Northwestern Polytechnical University Press, 1989: 24-50, 221-226, 296-304. (in Chinese)
[2]	彭吉. 直升机机载火控系统误差和精度相关性分析[J]. 现代制造技术与装备, 2019(11): 108-109.
	PENG J. Error and accuracy relativity analysis of helicopter airborne fire control system[J]. Modern Manufacturing Technology and Equipment, 2019(11): 108-109. (in Chinese)
[3]	曹林平, 黄长强. 武装直升机无控武器射击误差源分析[J]. 火力与指挥控制, 2000, 25(1): 70-73.
	CAO L P, HUANG C Q. Analysis for error source causing fire error of non-control weapon of armed heli-copter[J]. Fire Control and Command Control, 2000, 25(1): 70-73. (in Chinese)
[4]	罗鹏程, 傅攀峰. 武器装备敏感性分析方法综述[J]. 计算机工程与设计, 2008(21): 5546-5549.
	LUO P C, FU P F. Review on weapons and equipment sensitivity analysis methods[J]. Computer Engineering and Design, 2008(21): 5546-5549. (in Chinese)
[5]	李波, 雒浩然, 田琳宇, 等. 基于DBN效能拟合的舰艇编队作战效能敏感性分析[J]. 航空学报, 2019, 40(12): 323214. doi: 10.7527/S1000-6893.2019.23214
	LI B, LUO H R, TIAN L Y, et al. Sensitivity analysis of ship formation operational effectiveness based on DBN effectiveness fitting[J]. Acta Aeronautica et Astronautica Sinca, 2019, 40(12): 323214. (in Chinese)
[6]	李睿. Sobol’灵敏度分析方法在结构动态特性分析中的应用研究[D]. 长沙: 湖南大学, 2003.
	LI R. Application of Sobol’ sensitivity analysis method in structural dynamic characteristics analysis[D]. Changsha: Hunan University, 2003. (in Chinese)
[7]	张庆军, 谢颖, 吴松. 基于正交实验与回归分析的机载共形天线吸波涂层厚度预测[J]. 兵工学报, 2021, 42(12): 2693-2699. doi: 10.3969/j.issn.1000-1093.2021.12.017
	ZHANG Q J, XIE Y, WU S. Prediction of wave-absorbing coating thickness of airborne conformal antenna based on orthogonal experimental design and regression analysis[J]. Acta Armamentarii, 2021, 42(12): 2693-2699. (in Chinese)
[8]	贾晶晶, 张治民, 于建民, 等. 基于响应面法的轻质尾翼均匀挤压成形数值模拟及模具结构优化[J]. 兵工学报, 2024, 45(6): 1824-1839. doi: 10.12382/bgxb.2023.0187
	JIA J J, ZHANG Z M, YU J M, et al. Numerical simulation of uniform extrusion forming and die structure optimization of lightweight empennage-shape component based on response surface method[J]. Acta Armamentarii, 2024, 45(6): 1824-1839. (in Chinese)
[9]	FREY H C, PATIL S R. Identification and review of sensitivity analysis methods[J]. Risk Analysis, 2010, 22(3): 553-578.
[10]	CUKIER R I, SCHAIBLY J H, SHULER K E. Study of the sensitivity of coupled reaction systems to uncertainties in rate coefficients. III. analysis of the approximations[J]. Journal of Chemical Physics, 1975, 63(3): 1140-1149.
[11]	SOBOL I M. Sensitivity estimates for nonlinear mathematical models[J]. Matematicheskoe Modelirovanie, 1990, 2(1): 112-118.(in Russian)
[12]	胡子剑, 高晓光, 贺楚超, 等. 机载火控系统误差对精度影响的分析方法[J]. 航空兵器, 2018(5): 47-53.
	HU Z J, GAO X G, HE C C, et al. Analysis methods of the impact on the accuracy of airborne fire control system’s error[J]. Aero Weaponry, 2018(5): 47-53. (in Chinese)
[13]	尹文进, 张静远, 饶喆, 等. 基于Sobol指数法作战能力全局敏感性分析方法[J]. 航电技术, 2015, 35(12): 19-21.
	YIN W J, ZHANG J Y, RAO Z, et al. Global sensitivity analysis method based on Sobol index method[J]. Marine Electric, 2015, 35(12): 19-21. (in Chinese)
[14]	周经伦, 傅攀峰, 罗鹏程, 等. 基于方差的全局敏感性方法在空战效能分析中的运用[J]. 现代防御技术, 2007, 35(6): 22-27.
	ZHOU J L, FU P F, LUO P C, et al. Variance-based global sensitivity analysis method using in air combat effectiveness analysis[J]. Modern Defence Technology, 2007, 35(6): 22-27. (in Chinese)
[15]	贺楚超, 高晓光. 直升机火控系统精度敏感性分析的BNSobol法[J]. 航空学报, 2016, 37(10): 3110-3120. doi: 10.7527/S1000-6893.2016.0151
	HE C C, GAO X G. Sensitivity analysis on accuracy of helicopter fire control system: a BNSobol method[J]. Acta Aeronautica et Astronautica Sinca, 2016, 37(10): 3110-3120. (in Chinese)
[16]	PARUGGIA M. Sensitivity analysis in practice: a guide to assessing scientific models[J]. Journal of the American Statistical Association, 2012, 101(473): 398-399.
[17]	陈池礼. 针对某型直升机旋翼转速告警误差较大问题的分析[J]. 航空维修与工程, 2018 (4): 51-53.
	CHEN C L. Error analysis on a helicopter rotor speed alarm[J]. Aviation Maintenance and Engineering, 2018 (4): 51-53. (in Chinese)
[18]	王欣, 张凤阁, 姚俊, 等. 蒙特卡罗法分析随机风对火箭弹落点散布的影响[J]. 弹箭与制导学报, 2009, 29(1): 198-201.
	WANG X, ZHANG F G, YAO J, et al. Analyze effect of random wind on shell distribution of rocket using monte-carlo method[J]. Journal of Projectiles, Rockets, Missiles and Guidance, 2009, 29(1): 198-201. (in Chinese)
[19]	魏诗卉, 王明海. 导弹作战使用命中精度CEP评定方法研究[J]. 飞行力学, 2005, 23(4): 52-55.
	WEI S H, WANG M H. Research on the missile operational use cep assessment method[J]. Flight Dynamic, 2005, 23(4): 52-55. (in Chinese)
[20]	ASCOUGH J C, GREEN T R, MA L, et al. Key criteria and selection of sensitivity analysis methods applied to natural resource models[C]// Proceedings of International Congress on Modelling and Simulation. Melbourne, Australia: Modelling and Simulation Society of Australia and New Zealand, 2005: 2463-2469.
[21]	SALTELLI A, RATTO M, TARANTOLA S, et al. Sensitivity analysis practices: strategies for model-based inference[J]. Reliablity Engineering and System Safety, 2006, 91(10/11): 1109-1125.
[22]	蔡毅, 邢岩, 胡丹. 敏感性分析综述[J]. 北京师范大学学报(自然科学版), 2008, 44(1): 9-16.
	CAI Y, XING D, HU D. On sensitivity analysis[J]. Journal of Beijing Normal University (Natural Science), 2008, 44(1): 9-16. (in Chinese)
[23]	GLIMM B, HORROCKS I, LUTZ C, et al. Conjunctive query answering for the description logic SHIQ[J]. Journal of Artificial Intelligence Research, 2008, 31(3): 399-404.
[24]	HINTON G E, OSINDERO S, THE Y W. A fast learning algorithm for deep belief nets[J]. Neural Computing, 2006, 18(7): 1527-1554.
[25]	HINTON G E, SALAKHUTDINOV R R. Reducing the dimensionality of data with neural networks[J]. Science, 2019, 313(5786): 504-507.
[26]	高晓光. 航空军用飞行器导论[M]. 西安: 西北工业大学出版社, 2022: 93-115.
	GAO X G. Introduction to aviation military aircraft[M]. Xi’an: Northwestern Polytechnical University Press, 2022: 93-115. (in Chinese)
[27]	苏建刚, 杨若璋, 余文广, 等. 武器装备效能评估指标体系研究[C]// 第18届中国系统仿真技术及其应用学术年会论文集. 兰州: 中国自动化学会, 2017: 1-4.
	SU J G, YANG R Z, YU W G, et al. Research on effectiveness evaluation index system of weapon equipment[C]// Proceedings of the 18th Chinese Conference on System Simulation Technology and Its Application. Lanzhou: Chinese Society of Automation, 2017: 1-4. (in Chinese)
[28]	高晓光, 万开方, 李波, 等. 空战仿真模型及作战效能分析[M]. 北京: 国防工业出版社, 2022.
	GAO X G, WAN K F, LI B, et al. Air combat simulation model and combat effectiveness analysis[M]. Beijing: National Defense Industry Press, 2022. (in Chinese)

指标类型	参数	数值
	风速/(m·s^-1)	10
	风向角/(°)	30
环境	随机风场风速均方差/(m·s^-1)	3
	随机风场变化周期/s	1
	仿真步长/s	0.1
	仿真最长时间/s	60
	作战飞机高度/m	3000
作战飞机	作战飞机速度/(m·s^-1)	100
	作战飞机偏航角/(°)	15
	作战飞机俯仰角/(°)	0
	目标距离作战飞机距离/m	5000
目标	目标移动速度/(m·s^-1)	20
	目标半径/m	3
武器	武器杀伤半径/m	1

指标类型	参数	数值
	风速/(m·s^-1)	10
	风向角/(°)	30
环境	随机风场风速均方差/(m·s^-1)	3
	随机风场变化周期/s	1
	仿真步长/s	0.1
	仿真最长时间/s	60
	作战飞机高度/m	3000
作战飞机	作战飞机速度/(m·s^-1)	100
	作战飞机偏航角/(°)	15
	作战飞机俯仰角/(°)	0
	目标距离作战飞机距离/m	5000
目标	目标移动速度/(m·s^-1)	20
	目标半径/m	3
武器	武器杀伤半径/m	1

指标类型	参数	参数含义	数值
惯导测	x₁/m	惯导X轴方位测位误差	0~3
位精度	x₂/m	惯导Y轴方位测位误差	0~3
	x₃/m	惯导Z轴方位测位误	0~3
惯导测	x₄/(m·s^-1)	惯导X轴方位测速误差	0~3
速精度	x₅/(m·s^-1)	惯导Y轴方位测速误差	0~3
	x₆/(m·s^-1)	惯导Z轴方位测速误差	0~3
惯导测	x₇/(°)	惯导探测姿态误差	0~0.2
角精度	x₈/(°)	惯导探测俯仰误差	0~0.2
大气数据系统精度	x₉/m	大气机测高误差	20
雷达测距精度	x₁₀/m	雷达探测距离误差	0~3
雷达测	x₁₁/(°)	雷达探测方位角误差	0~0.2
角精度	x₁₂/(°)	雷达探测俯仰角误差	0~0.2

指标类型	参数	参数含义	数值
惯导测	x₁/m	惯导X轴方位测位误差	0~3
位精度	x₂/m	惯导Y轴方位测位误差	0~3
	x₃/m	惯导Z轴方位测位误	0~3
惯导测	x₄/(m·s^-1)	惯导X轴方位测速误差	0~3
速精度	x₅/(m·s^-1)	惯导Y轴方位测速误差	0~3
	x₆/(m·s^-1)	惯导Z轴方位测速误差	0~3
惯导测	x₇/(°)	惯导探测姿态误差	0~0.2
角精度	x₈/(°)	惯导探测俯仰误差	0~0.2
大气数据系统精度	x₉/m	大气机测高误差	20
雷达测距精度	x₁₀/m	雷达探测距离误差	0~3
雷达测	x₁₁/(°)	雷达探测方位角误差	0~0.2
角精度	x₁₂/(°)	雷达探测俯仰角误差	0~0.2

指标	Sobol算法^[13]		BNSobol算法^[15]		GSADBN算法		FAST算法^[10]
指标	主效应值	排序	主效应值	排序	主效应值	排序	主效应值	排序
x₁	0.050	11	0.039	10	0.012	10	0.016	9
x₂	0.055	9	0.043	9	0	11	0.057	3
x₃	0.066	7	0.028	12	0.097	5	0.024	7
x₄	0.168	1	0.055	6	0.368	1	0.509	1
x₅	0.103	4	0.047	8	0.127	2	0.209	2
x₆	0.117	2	0.083	5	0	11	0.052	4
x₇	0.066	7	0.161	1	0.014	9	0.012	11
x₈	0.056	8	0.156	2	0.139	4	0.020	8
x₉	0.116	3	0.036	11	0.170	3	0.011	12
x₁₀	0.077	5	0.049	7	0.024	7	0.050	5
x₁₁	0.051	10	0.155	3	0.021	8	0.025	6
x₁₂	0.076	6	0.146	4	0.027	6	0.015	10