基于高斯模型和RMSD-DS的火炮反后坐装置故障诊断方法

doi:10.12382/bgxb.2022.0635

摘要/Abstract

摘要：

针对变工况作战环境下火炮反后坐装置状态信号的模糊性和高冲突性等特点所导致的故障诊断可信性不高的问题,提出一种基于高斯模型和均方根偏移(RMSD)-Dempster-Shafer(DS)故障诊断方法,实现信号的模糊化定量表征以及高度冲突证据融合的高可信性诊断。采用高斯模型求解火炮反后坐装置的多个故障特性信号所对应证据的基本概率分配,实现对信号模糊性的定量表征处理;通过构造RMSD-DS冲突系数求解证据间相似度,依据相似度定义各证据重要度,实现每个证据在融合过程中重要程度的定量描述;提出依靠证据重要度的权重求解及证据融合方法,实现高度冲突证据的可靠融合。仿真试验数据应用表明,所提方法相较于其他代表性方法具有更高的准确性和鲁棒性。

关键词: 高斯模型, 火炮反后坐装置, 证据理论, 故障诊断

Abstract:

To deal with the problem of low reliability of fault diagnosis caused by the ambiguity and high conflict of the state signal of the anti-recoil device of the gun under variable operating conditions, a fault diagnosis method based on the Gaussian model and RMSD (Root Mean Square Deviation)-DS (Dempster-Shafer) is proposed to realize the quantitative characterization of signal fuzziness and the high reliability diagnosis of high-conflict evidence fusion. This method uses the Gaussian model to solve the basic probability assignment of the evidence corresponding to multiple fault feature signals of the anti-recoil device, thus realizing the quantitative characterization of the signal ambiguity. By constructing the RMSD-DS conflict coefficient to solve the similarity between evidences, the importance of each evidence is defined according to the similarity, and the quantitative description of the importance of each evidence in the fusion process is realized. The weight solution and evidence fusion method based on the importance of evidence are put forward to realize the reliable fusion of highly conflicting evidence. The application of simulation test data shows that the proposed method has higher accuracy and robustness than other representative methods.

Key words: Gaussian model, gun’s anti-recoil device, evidence theory, fault diagnosis

中图分类号:

E924
TJ303.4

魏剑峰, 张发平, 卢继平, 杨向飞, 杨鹏楷. 基于高斯模型和RMSD-DS的火炮反后坐装置故障诊断方法[J]. 兵工学报, 2023, 44(10): 3101-3114.

WEI Jianfeng, ZHANG Faping, LU Jiping, YANG Xiangfei, YANG Pengkai. Fault Diagnosis for Gun’s Anti-recoil Device Based on Gaussian Model and RMSD-DS[J]. Acta Armamentarii, 2023, 44(10): 3101-3114.

图/表 16

图1 火炮简易结构图

Fig.1 Simple structure of a gun

图2 后坐部分受力情况(后坐运动)

Fig.2 Force analysis of the recoil part (recoiling)

图3 后坐部分受力情况(复进运动)

Fig.3 Force analysis of the recoil part (counter-recoiling)

图4 后坐和复进运动过程中位移与速度变化

Fig.4 Displacement and velocity curves during recoiling and counter-recoiling motion

图5 不同状态下后坐复进运动中位移和速度变化

Fig.5 Displacement and velocity curves during recoiling and counter-recoiling motion in different states

图6 故障诊断方法流程图

Fig.6 Flow chart of the fault diagnosis method

图7 不同故障特性信号上的匹配程度

Fig.7 Matching degree on signals with different fault characteristics

表1 不同方法融合结果

Table 1 Fusion results of different methods

方法	m₁, m₂	m₁, m₂, m₃	m₁, m₂, m₃, m₄
DS	m(A)=0 m(B)=0.8571 m(C)=0.1429	m(A)=0 m(B)=0.6316 m(C)=0.3684	m(A)=0 m(B)=0.3288 m(C)=0.6712
Yager^[13]	m(A)=0 m(B)=0.1800 m(C)=0.0300 m(O)=0.7900	m(A)=0 m(B)=0.0180 m(C)=0.0105 m(O)=0.9715	m(A)=0 m(B)=0.0018 m(C)=0.0037 m(O)=0.9945
Murphy^[16]	m(A)=0.1543 m(B)=0.7469 m(C)=0.0998	m(A)=0.3500 m(B)=0.5224 m(C)=0.1276	m(A)=0.6027 m(B)=0.2627 m(C)=0.1346
RMSD-DS	m(A)=0.1543 m(B)=0.7469 m(C)=0.0998	m(A)=0.5100 m(B)=0.3280 m(C)=0.1620	m(A)=0.8493 m(B)=0.0114 m(C)=0.1393

图8 不同方法的融合结果

Fig.8 Fusion results of different methods

图9 不同故障状态所对应的故障特征信号数据

Fig.9 Fault feature signal corresponding to different fault states

表2 高斯模型的平均值和标准差

Table 2 Mean and standard deviations of the Gaussian model

故障状态	$μ X m a x$ /m	$σ X m a x$ /m	$μ v m a x$ / (m·s^-1)	$σ v m a x$ / (m·s^-1)	$μ U m a x$ / (m·s^-1)	$σ U m a x$ / (m·s^-1)	$μ U e n d$ / (m·s^-1)	$σ U e n d$ / (m·s^-1)
WTR	1.0224	0.0371	10.6842	0.0642	2.2759	0.1264	0.1388	0.0055
GLR	0.9595	0.0108	10.5730	3×10^-4	2.0598	0.0029	0.1521	0.0020
BRPW	0.9781	0.0108	10.5876	0.0086	1.7774	0.1675	0.1088	0.0216

表2 高斯模型的平均值和标准差

Table 2 Mean and standard deviations of the Gaussian model

故障状态	$μ X m a x$ /m	$σ X m a x$ /m	$μ v m a x$ / (m·s^-1)	$σ v m a x$ / (m·s^-1)	$μ U m a x$ / (m·s^-1)	$σ U m a x$ / (m·s^-1)	$μ U e n d$ / (m·s^-1)	$σ U e n d$ / (m·s^-1)
WTR	1.0224	0.0371	10.6842	0.0642	2.2759	0.1264	0.1388	0.0055
GLR	0.9595	0.0108	10.5730	3×10^-4	2.0598	0.0029	0.1521	0.0020
BRPW	0.9781	0.0108	10.5876	0.0086	1.7774	0.1675	0.1088	0.0216

图10 不同故障特征信号上所属的高斯模型

Fig.10 Gaussian models belonging to different fault feature signals

图11 不同方法的诊断结果(混淆矩阵)

Fig.11 Diagnostic results of different methods (confusion matrix)

表3 不同方法的故障诊断正确率

Table 3 Accuracy of fault diagnosis of different methods%

方法	WTR	GLR	BRPW	平均值
DS	91.5	1.5	95.5	62.8
Yager^[13]	53.5	0	0	17.8
Murphy^[16]	91.0	89.0	96.0	92.0
BP^[4]	99.5	17.5	100	72.3
RMSD-DS	90.5	95.5	97.5	94.5

图12 不同方法的对比

Fig.12 Comparison of different methods

图13 不同方法的ROC曲线

Fig.13 ROC curves of the five methods

参考文献 22

[1]	张相炎, 郑建国, 袁人枢. 火炮设计理论[M]. 北京: 北京理工大学出版社, 2014.
	ZHANG X Y, ZHENG J G, YUAN R S. Theory of artillery gun design[M]. Beijing: Beijing Institute of Technology Press, 2014. (in Chinese)
[2]	杜中华, 狄长春. 某火炮复杂反后坐装置工作特性仿真分析[J]. 机械工程师, 2011(2): 96-99.
	DU Z H, DI C H. Working characteristic simulation analysis of a gun with complex recoil system[J]. Mechanical Engineer, 2011(2): 96-99. (in Chinese)
[3]	高跃飞. 火炮反后坐装置设计[M]. 北京: 国防工业出版社, 2015.
	GAO Y F. Artillery gun anti-recoil device design[M]. Beijing: National Defense Industry Press, 2015. (in Chinese)
[4]	张晓东, 傅建平, 张培林. 基于BP神经网络的火炮反后坐装置故障诊断[J]. 兵工自动化, 2006(8): 15-16,20.
	ZHANG X D, FU J P, ZHANG P L. Fault diagnosis of gun anti-recoil mechanism based on BP neural networks[J]. Ordnance Industry Automation, 2006(8): 15-16,20. (in Chinese)
[5]	曹亭. 火炮状态诊断与应急处理方法研究[D]. 南京: 南京理工大学, 2013.
	CAO T. Research on state-diagnosis and emergency treatment of artillery gun[D]. Nanjing: Nanjing University of Science and Technology, 2013. (in Chinese)
[6]	JIANG W, HU W W, XIE C H. A new engine fault diagnosis method based on multi-sensor data fusion[J]. Applied Sciences, 2017, 7(3): 1-18. doi: 10.3390/app7010001 URL
[7]	尹东亮, 黄晓颖, 吴艳杰, 等. 基于云模型和改进D-S证据理论的目标识别决策方法[J]. 航空学报, 2021, 42(12): 299-310.
	YIN D L, HUANG X Y, WU Y J, et al. Research on target identification decision method based on cloud model and improved D-S evidence theory[J]. Acta Aeronautica et Astronautica Sinica, 2021, 42(12): 299-310. (in Chinese)
[8]	吴斌, 衣晓, 李双明. 一种区间距离的冲突证据组合方法[J]. 兵工学报, 2020, 41(6): 1140-1150. (in Chinese) doi: 10.3969/j.issn.1000-1093.2020.06.010
	WU B, YI X, LI S M. Conflict evidence combination method based on interval distance[J]. Acta Armamentarii, 2020, 41(6): 1140-1150. (in Chinese) doi: 10.3969/j.issn.1000-1093.2020.06.010
[9]	XIAO F Y. Multi-sensor data fusion based on the belief divergence measure of evidences and the belief entropy[J]. Information Fusion, 2019, 46: 23-32. doi: 10.1016/j.inffus.2018.04.003 URL
[10]	ZHAO R Y, LI C L, TIAN X K. A novel industrial multimedia: rough set based fault diagnosis system used in CNC grinding machine[J]. Multimedia Tools and Applications, 2017, 76(191): 19913-19926. doi: 10.1007/s11042-016-3878-0 URL
[11]	LIU J G, HUANG J, SUN R, et al. Data fusion for multi-source sensors using GA-PSO-BP neural network[J]. IEEE Transactions on Intelligent Transportation Systems, 2021, 22(10): 6583-6598. doi: 10.1109/TITS.2020.3010296 URL
[12]	MA W J, JIANG Y C, LUO X D. A flexible rule for evidential combination in Dempster-Shafer theory of evidence[J]. Applied Soft Computing, 2019, 85: 105512. doi: 10.1016/j.asoc.2019.105512 URL
[13]	YAGER R R. On the Dempster-shafer framework and new combination rules[J]. Information Sciences, 1987, 41(2): 93-137. doi: 10.1016/0020-0255(87)90007-7 URL
[14]	张昌芳, 杨宏文, 胡卫东, 等. 基于改进冲突度量的证据组合新规则[J]. 系统工程与电子技术, 2009, 31(11): 2637-2640,2727.
	ZHANG C F, YANG H W, HU W D, et al. New rule of evidence combination based on improved conflict measure[J]. Systems Engineering and Electronics, 2009, 31(11): 2637-2640,2727. (in Chinese)
[15]	孙贵东, 关欣, 衣晓, 等. 互补证据组合规则[J]. 电子学报, 2018, 46(11): 2738-2745. doi: 10.3969/j.issn.0372-2112.2018.11.022
	SUN G D, GUAN X, YI X, et al. Complementary combination rule for evidence[J]. Acta Electronica Sinica, 2018, 46(11): 2738-2745. (in Chinese)
[16]	MURPHY C K. Combining belief functions when evidence conflicts[J]. Decision Support Systems, 2000, 29(1): 1-9. doi: 10.1016/S0167-9236(99)00084-6 URL
[17]	TANG X H, GU X, WANG J C, et al. A bearing fault diagnosis method based on feature selection feedback network and improved DS evidence fusion[J]. IEEE Access, 2020, 8: 20523-20536. doi: 10.1109/Access.6287639 URL
[18]	ZHANG W Q, DENG Y. Combining conflicting evidence using the DEMATEL method[J]. Soft Computing, 2019, 23: 8207-8216. doi: 10.1007/s00500-018-3455-8
[19]	张培林, 王成, 张晓东, 等. 火炮后坐复进运动仿真技术及应用[M]. 北京: 国防工业出版社, 2015.
	ZHANG P L, WANG C, ZHANG X D, et al. Simulation technique and application of gun recoiling and counter-recoiling movemnet[M]. Beijing: National Defense Industry Press, 2015. (in Chinese)
[20]	WANG T, QIAO Y H, DING W Z, et al. Improved fragment sampling for ab initio protein structure prediction using deep neural networks[J]. Nature Machine Intelligence, 2019, 1(8): 347-355. doi: 10.1038/s42256-019-0075-7
[21]	梁永亮, 郭汉琮, 薛永端. 基于特征气体关联特征的变压器故障诊断方法[J]. 高电压技术, 2019, 45(2): 386-392.
	LIANG Y L, GUO H Z, XUE Y D. Transformer fault diagnosis method based on association characteristics of characteristic gases[J]. High Voltage Engineering, 2019, 45(2): 386-392. (in Chinese)
[22]	LI L, ZHU F, SUN H, et al. Multi-source information fusion and deep-learning-based characteristics measurement for exploring the effects of peer engagement on stock price synchronicity[J]. Information Fusion, 2021, 69: 1-21. doi: 10.1016/j.inffus.2020.11.006 URL