基于多特征融合的电磁换向阀故障模式识别

马栋; 刘志浩; 高钦和; 黄通

doi:10.13700/j.bh.1001-5965.2021.0367

基于多特征融合的电磁换向阀故障模式识别

doi: 10.13700/j.bh.1001-5965.2021.0367

火箭军工程大学兵器发射理论与技术国家重点学科实验室，西安 710025

基金项目: 国家自然科学基金(51905541)；陕西省自然科学基础研究计划(2020JQ487)；陕西省高校科协青年人才托举计划(20190412)

详细信息

通讯作者:
E-mail：liuzh_epgc@163.com

中图分类号: TH137.52⁺3
计量
- 文章访问数: 284
- HTML全文浏览量: 74
- PDF下载量: 35
- 被引次数: 0
出版历程
- 收稿日期: 2021-07-01
- 录用日期: 2021-11-01
- 网络出版日期: 2021-11-26
- 整期出版日期: 2023-04-30

Solenoid directional control valve fault pattern recognition based on multi-feature fusion

National Key Discipline Laboratory of Armament Launch Theory & Technology，Rocket Force University of Engineering，Xi’an 710025，China

Funds: National Natural Science Foundation of China (51905541); Natural Science Basic Research Program of Shaanxi (2020JQ487); Young Talent Fund of University Association for Science and Technology in Shaanxi, China (20190412)

More Information

Corresponding author: E-mail：liuzh_epgc@163.com

摘要

摘要:
为提高基于驱动端电流检测的电磁换向阀故障诊断方法的可靠性和识别准确度，开展了电磁换向阀故障模式识别方法研究。提出一种基于多特征融合的方法对电流信号时频分析和时域参数的特征值提取融合；通过设计电磁换向阀驱动端电流信号的采集实验，获取电磁换向阀驱动端电流的时域信号和二阶变化率的多特征曲线，提取时域参数及二阶变化率相应频带能量作为特征值，构建多特征融合的特征向量；采用基于径向基核函数的多分类支持向量机对电磁换向阀进行模式识别。结果表明：基于多特征融合的支持向量机较基于能量特征值的支持向量机可提升8.7%的识别精度和42.11%的验证准确率。
- 电磁阀 /
- 故障诊断 /
- 电流检测 /
- 小波包分解 /
- 主元分析 /
- 多特征融合 /
- C参数支持向量机
Abstract:
In order to further improve the reliability and recognition accuracy of the solenoid valve fault diagnosis method based on current detection at the drive end, a research was conducted on the solenoid valve fault pattern recognition method. First, a method for extracting eigenvalues based on time-frequency analysis of current signals and time-domain parameters was proposed; then, through designing an acquisition experiment of the current signal at the solenoid valve drive end, the time domain signal of the solenoid valve drive end current and the multi-characteristic curve of the second-order rate of change were obtained. Meanwhile, the time-domain parameters and the frequency band energy corresponding to the second-order rate of change were extracted as the characteristic value, in order to construct the feature vector of multi-feature fusion. Finally, a multi-class support vector machine based on the radial basis kernel function was used to identify the electromagnetic directional valve pattern. The research results showed that compared with the support vector machine based on energy eigenvalues, the support vector machine based on multi-feature fusion can improve the recognition accuracy by 8.7% and the verification accuracy by 42.11%.
- solenoid valve /
- fault diagnosis /
- current detection /
- wavelet packet decomposition /
- principal component analysis /
- multi-feature fusion /
- C-support vector machine

HTML全文

图 1 小波包提取特征值步骤

Figure 1. Wavelet packet extraction extraction step

下载: 全尺寸图片幻灯片

图 2 三层小波包分解示意图

注：S为原始信号；A为高频；D为低频；数字为分解的层数。

Figure 2. Schematic diagram of three-layer wavelet packet decomposition

下载: 全尺寸图片幻灯片

图 3 电磁换向阀故障诊断流程

Figure 3. Solenoid directional control valve fault diagnosis process

下载: 全尺寸图片幻灯片

图 4 4WE10E31B/CG24N9Z5L型电磁换向阀

Figure 4. 4WE10E31B/CG24N9Z5L solenoid directional control valve

下载: 全尺寸图片幻灯片

图 5 驱动端电流检测系统

Figure 5. Drive end current detection system

下载: 全尺寸图片幻灯片

图 6 驱动端电流和阀芯位移

Figure 6. Drive end current and spool displacement

下载: 全尺寸图片幻灯片

图 7 不同状态下电磁换向阀驱动端电流

Figure 7. Solenoid directional control valve drive end current in different states

下载: 全尺寸图片幻灯片

图 8 不同状态下电流信号二阶变化率

Figure 8. Second-order guide of current signal in different states

下载: 全尺寸图片幻灯片

图 9 小波重构信号

Figure 9. Wavelet reconstruction signal

下载: 全尺寸图片幻灯片

图 10 不同特征向量训练集预测结果

Figure 10. Prediction results of different feature vector training sets

下载: 全尺寸图片幻灯片

图 11 不同特征向量测试集预测结果

Figure 11. Prediction results of different feature vector test sets

下载: 全尺寸图片幻灯片

表 1 不同SVM准确率对比

Table 1. Comporison of different SVM accuracy rates

SVM类型	训练准确率/%	测试准确率/%	五折交叉验证准确率/%
C-SVC	100	91.30	50
υ-SVC	94.74	84.78	50
ε-SVR	100	86.96
υ-SVR	100	86.96

下载: 导出CSV

表 2 不同核函数SVM准确率对比

Table 2. Comporison of SVM accuracy rates of different kernel functions

核函数	训练准确率/%	测试准确率/%	五折交叉验证准确率/%
Polynomial	97.74	89.13	50
RBF	100	91.30	50
Sigmoid	76.32	67.39	42.11

下载: 导出CSV

表 3 四种不同状况特征向量

Table 3. Four feature vectors under different situations

电磁换向阀状态	组号	E₃₀	E₃₁	E₃₂	E₃₃	E₃₄	E₃₅	E₃₆	E₃₇
阀芯正常	第1组	0.0871	0.1096	0.0829	0.1920	0.2944	0.6247	0.1529	0.6606
阀芯正常	第2组	0.0769	0.1060	0.0997	0.2109	0.2433	0.6577	0.1945	0.6315
弹簧断裂	第1组	0.0121	0.0302	0.0969	0.0845	0.6050	0.5377	0.2156	0.5299
弹簧断裂	第2组	0.0013	0.0034	0.1421	0.1759	0.6216	0.7014	0.1777	0.1972
阀芯卡死	第1组	0.0219	0.0747	0.1492	0.1535	0.6368	0.4306	0.3047	0.5142
阀芯卡死	第2组	0.0226	0.0388	0.0892	0.0574	0.5849	0.4790	0.2479	0.5947
阀芯轻微卡滞	第1组	0.0326	0.1608	0.1768	0.1104	0.2458	0.4869	0.2196	0.7642
阀芯轻微卡滞	第2组	0.0400	0.0893	0.2242	0.1792	0.3224	0.7458	0.2200	0.4467

下载: 导出CSV

表 4 各主元贡献率

Table 4. Contribution rate of each principal

主元数	贡献率/%
1	60.0411
2	29.1488
3	9.0540
4	1.6772
5	0.0659
6	0.0071
7	0.0051
8	0.0005
9	0.0002
10	0.0001
11	1.21×10⁻⁵

下载: 导出CSV

表 5 前4主元系数

Table 5. First 4 pivot coefficients

时域信号	第1主元	第2主元	第3主元	第4主元
p₁	−0.3333	0.2689	0.171	0.1703
p₂	−0.2795	−0.330	−0.290	0.5124
p₃	−0.2594	0.414	0.0356	0.1565
p₄	0.2871	0.2245	0.5244	−0.3011
p₅	−0.3333	0.2687	0.1719	0.1717
p₆	0.2469	0.3678	−0.4031	−0.0773
p₇	0.3537	0.0989	0.3041	0.520
p₈	0.3529	0.2034	−0.0267	0.4869
p₉	−0.350	0.1216	0.3778	0.0651
p₁₀	0.3101	−0.3042	0.2449	0.2126
p₁₁	0.1439	0.4806	−0.3499	0.0305

下载: 导出CSV

参考文献(16)

[1]	刘志浩, 高钦和, 牛海龙, 等. 基于驱动端电流检测的电磁阀故障诊断研究[J]. 兵工学报, 2014, 35(7): 1083-1090. doi: 10.3969/j.issn.1000-1093.2014.07.023 LIU Z H, GAO Q H, NIU H L, et al. The fault diagnosis of electromagnetic valves based on driving current detection[J]. Acta Armamentarii, 2014, 35(7): 1083-1090(in Chinese). doi: 10.3969/j.issn.1000-1093.2014.07.023
[2]	蔡伟, 郑贤林, 张志利, 等. 液压电磁阀故障机理分析与瞬态特性仿真[J]. 仪器仪表学报, 2011, 32(12): 2726-2733. doi: 10.19650/j.cnki.cjsi.2011.12.014 CAI W, ZHENG X L, ZHANG Z L, et al. Failure mechanism analysis and transient characteristic simulation of hydraulic electromagnetic valve[J]. Chinese Journal of Scientific Instrument, 2011, 32(12): 2726-2733(in Chinese). doi: 10.19650/j.cnki.cjsi.2011.12.014
[3]	孙昱, 何林. 基于电机电流信号的齿轮泵故障识别方法[J]. 机床与液压, 2021, 49(17): 191-195. doi: 10.3969/j.issn.1001-3881.2021.17.036 SUN Y, HE L. Fault identification method of gear pump based on motor current signal[J]. Machine Tool & Hydraulics, 2021, 49(17): 191-195(in Chinese). doi: 10.3969/j.issn.1001-3881.2021.17.036
[4]	刘洋, 刘晓波, 梁珊. 基于傅里叶分解方法的航空发动机转子故障诊断[J]. 中国机械工程, 2019, 30(18): 2156-2163. LIU Y, LIU X B, LIANG S. Aeroengine rotor fault diagnosis based on Fourier decomposition method[J]. China Mechanical Engineering, 2019, 30(18): 2156-2163(in Chinese).
[5]	尹欣, 王雯. 基于小波包“能量-故障”的高温合金焊接缺陷特征值提取[J]. 煤矿机械, 2008, 29(11): 188-189. doi: 10.3969/j.issn.1003-0794.2008.11.084 YIN X, WANG W. Defect features extraction of heat metal alloys based on “energy-fault” of wavelet packet[J]. Coal Mine Machinery, 2008, 29(11): 188-189(in Chinese). doi: 10.3969/j.issn.1003-0794.2008.11.084
[6]	张雪英, 任永梅, 贾海蓉. 用后验信噪比修正小波包自适应阈值的语音增强算法[J]. 中南大学学报(自然科学版), 2013, 44(11): 4566-4573. ZHANG X Y, REN Y M, JIA H R. Speech enhancement algorithm based on wavelet packet adaptive threshold revised by posteriori SNR[J]. Journal of Central South University (Science and Technology), 2013, 44(11): 4566-4573(in Chinese).
[7]	BRO R, SMILDE A K. Principal component analysis[J]. Analytical Methods, 2014, 6(9): 2812-2831. doi: 10.1039/C3AY41907J
[8]	SAINANI K L. Introduction to principal components analysis[J]. PM & R, 2014, 6(3): 275-278.
[9]	LI L M, ZHAO J, WANG C R, et al. Comprehensive evaluation of robotic global performance based on modified principal component analysis[J]. International Journal of Advanced Robotic Systems, 2020, 17(4): 172988141989688.
[10]	WANG L H, WU X Q, ZHANG C Y, et al. Hydraulic system fault diagnosis method based on a multi-feature fusion support vector machine[J]. The Journal of Engineering, 2019, 2019(13): 215-218. doi: 10.1049/joe.2018.9028
[11]	徐玉秀, 杨文平, 吕轩, 等. 基于支持向量机的汽车发动机故障诊断研究[J]. 振动与冲击, 2013, 32(8): 143-146. doi: 10.3969/j.issn.1000-3835.2013.08.024 XU Y X, YANG W P, LYU X, et al. Fault diagnosis for a car engine based on support vector machine[J]. Journal of Vibration and Shock, 2013, 32(8): 143-146(in Chinese). doi: 10.3969/j.issn.1000-3835.2013.08.024
[12]	张学工. 关于统计学习理论与支持向量机[J]. 自动化学报, 2000, 26(1): 32-42. doi: 10.16383/j.aas.2000.01.005 ZHANG X G. Introduction to statistical learning theory and support vector machines[J]. Acta Automatica Sinica, 2000, 26(1): 32-42(in Chinese). doi: 10.16383/j.aas.2000.01.005
[13]	VAPNIK V N. An overview of statistical learning theory[J]. IEEE Transactions on Neural Networks, 1999, 10(5): 988-999. doi: 10.1109/72.788640
[14]	WESTON J, WATKINS C. Support vector machines for multi-class pattern recognition[C]// Proceedings of 7th European Symposium on Artificial Neural Networks, 1999: 219-224.
[15]	奉国和. SVM分类核函数及参数选择比较[J]. 计算机工程与应用, 2011, 47(3): 123-124. doi: 10.3778/j.issn.1002-8331.2011.03.037 FENG G H. Parameter optimizing for support vector machines classification[J]. Computer Engineering and Applications, 2011, 47(3): 123-124(in Chinese). doi: 10.3778/j.issn.1002-8331.2011.03.037
[16]	赵薇, 靳聪, 涂中文, 等. 基于多特征融合的SVM声学场景分类算法研究[J]. 北京理工大学学报, 2020, 40(1): 69-75. doi: 10.15918/j.tbit1001-0645.2018.171 ZHAO W, JIN C, TU Z W, et al. Support vector machine for acoustic scene classification algorithm research based on multi-features fusion[J]. Transactions of Beijing Institute of Technology, 2020, 40(1): 69-75(in Chinese). doi: 10.15918/j.tbit1001-0645.2018.171