-
摘要:
通过对航空发动机故障诊断，能够正确判断各部件工作状态，快速确定维修方案，保证飞行安全。在结合深度信念网络和决策融合算法的基础上，提出了基于深度学习的航空发动机故障融合诊断模型。该模型通过分析发动机的大量性能参数，先利用深度学习模型提取出性能参数中的隐藏特征，得出故障分类置信度；其后对多次故障分类结果进行决策融合，从而得出更准确的诊断结果。将普惠JT9D发动机故障系数用于数据仿真，通过算例验证本文算法的有效性；算例计算结果表明：多次实验结果经数据融合提高了可信度，该模型具有较高的故障分类诊断准确性和抗干扰能力。
Abstract:Through the fault diagnosis of aero-engine, the working status of each component can be correctly judged, and the maintenance program can be determined quickly to ensure the safety of flight. Based on the combination of deep belief network and decision fusion theory, the fault fusion diagnosis model of aero-engine based on deep learning was proposed. This model, through analyzing a large number of engine performance parameters, starts with getting fault classification confidence via hidden features in engine performance parameters extracted by deep learning algorithm, and then the multiple fault classification results were fused by decision fusion method to get more accurate results. The JT9D engine failure coefficient was simulated as data to prove the validity of the method. The results of an example show that the reliability of the data has been improved by fault fusion diagnosis of several experimental results, and the model has high fault classification and diagnosis accuracy and anti-interference ability.
-
Key words:
- deep learning /
- fault diagnosis /
- decision fusion /
- anti-interference ability /
- aero-engine
-
图 4 基于深度学习的航空发动机故障融合诊断流程
Figure 4. Fault fusion diagnosis folw chart for aero-engine based on deep learning
图 5 网络结构为[4, 8, 9, 10]的DBN网络重构误差图
Figure 5. DBN reconstruction error graph with network structure of [4, 8, 9, 10]
表 1 JT9D发动机故障
Table 1. Engine failure of JT9D
序号 发动机故障类型 性能参数偏差 ΔEGT/℃ ΔFF/% ΔN1/% ΔN2/% F01 风扇效率+1% -1.00 -0.25 -0.25 -0.05 F02 风扇流量+1% 4.50 0.80 -0.85 0.05 F03 低压压气机效率+1% -2.00 -0.15 0.10 -0.15 F04 低压压气机流量+1% -2.00 -0.40 -2.00 -0.05 F05 高压压气机效率+1% -7.00 -0.85 -0.10 0.10 F06 高压压气机流量+1% -0.05 -0.10 0 -0.25 F07 高压涡轮效率+1% -0.85 -1.05 -0.10 0.20 F08 低压涡轮效率+1% -4.50 0 0.45 0 F09 3.0+3.5放气活门开度-20% -11.20 -2.40 -0.04 -0.24 F10 第1级涡轮导向叶片面积+1% 2.50 0.35 0.05 -0.15 
下载: 导出CSV
表 2 DBN网络结构与重构误差
Table 2. DBN network structure and reconstruction error
DBN网络结构 重构误差 [4, 6, 10] 0.083 [4, 8, 10] 0.047 [4, 6, 8, 10] 0.076 [4, 8, 12, 10] 0.037 [4, 8, 9, 10] 0.028 [4, 8, 9, 12, 10] 0.049 [4, 6, 8, 12, 10] 0.065
下载: 导出CSV
表 3 故障分类置信度
Table 3. Confidence level of fault classification
故障类型 置信度1 置信度2 置信度3 置信度4 置信度5 F01 9.00×10-8 1.08×10-6 6.01×10-7 2.35×10-6 4.90×10-5 F02 2.41×10-7 7.09×10-7 5.39×10-7 3.77×10-7 6.06×10-7 F03 5.16×10-3 2.60×10-3 3.05×10-3 1.40×10-3 6.57×10-5 F04 6.30×10-5 2.35×10-4 1.81×10-4 1.15×10-4 2.50×10-5 F05 2.78×10-1 7.97×10-1 6.44×10-1 9.47×10-1 9.99×10-1 F06 1.16×10-5 2.05×10-6 3.78×10-6 3.17×10-7 1.91×10-9 F07 1.01×10-3 7.27×10-4 1.08×10-3 1.88×10-4 1.18×10-5 F08 1.10×10-5 4.65×10-5 2.30×10-5 8.12×10-5 4.31×10-4 F09 7.16×10-1 2.01×10-1 3.52×10-1 5.14×10-2 5.54×10-4 F10 4.66×10-7 4.68×10-7 4.34×10-7 1.40×10-7 3.39×10-8
下载: 导出CSV
表 4 故障诊断结果
Table 4. Results of fault diagnosis
故障类型 C(F) 诊断结果 F01 3.9×10-5 0 F02 7.6×10-7 0 F03 7.16×10-5 0 F04 0.38×10-3 0 F05 0.998 1 F06 4.32×10-9 0 F07 3.25×10-5 0 F08 0.161×10-3 0 F09 0.129 0 F10 4.14×10-8 0
下载: 导出CSV
表 5 故障诊断正确率对比
Table 5. Comparison of fault diagnosis accuracy
模型 正确率/% 基于深度学习的故障融合诊断模型 99.58 深度信念网络模型 95.30 BP神经网络模型 82.13
下载: 导出CSV
-
[1] 黄强, 王建, 张桂刚.一种航空发动机传感器故障诊断方法[J].传感技术学报, 2014, 27(10):1315-1320. doi: 10.3969/j.issn.1004-1699.2014.10.003HUANG Q, WANG J, ZHANG G G.A kind of approach for aero-engine sensor fault diagnosis[J].Chinese Journal of Sensors and Actuators, 2014, 27(10):1315-1320(in Chinese). doi: 10.3969/j.issn.1004-1699.2014.10.003 [2] 薛薇, 王涛.基于一组卡尔曼滤波器信息融合的故障诊断[J].计算机测量与控制, 2015, 23(7):2285-2287. http://d.old.wanfangdata.com.cn/Periodical/jsjzdclykz201507013XUE W, WANG T.Application of a bank of Kalman filters for aircraft engine sensor actuator fault diagnosis[J].Computer Measurement and Control, 2015, 23(7):2285-2287(in Chinese). http://d.old.wanfangdata.com.cn/Periodical/jsjzdclykz201507013 [3] 俞刚, 黄金泉.基于支持向量机和卡尔曼滤波器的航空发动机故障诊断研究[J].航空发动机, 2012, 38(1):47-50. http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=hkfdj201201015YU G, HUANG J Q.Aeroengine fault diagnosis based on support vector machine and Kalman filter[J].Aero-Engine, 2012, 38(1):47-50(in Chinese). http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=hkfdj201201015 [4] 廖洪一. 基于极限学习机的航空发动机故障诊断研究[D]. 广汉: 中国民用航空飞行学院, 2012.LIAO H Y. Research on aero-engine fault diagnosis based on extreme learning machine[D]. Guanghan: Civil Aviation Flight University of China, 2012(in Chinese). [5] 赵世荣, 黄向华.应用神经网络信息融合诊断航空发动机故障[J].航空动力学报, 2008, 23(1):164-168. http://d.wanfangdata.com.cn/Periodical_hkdlxb200801028.aspxZHAO S R, HUANG X H.Fault diagnosis for aeroengine gas path components based on neural network multisensor data fusion[J].Journal of Aerospace Power, 2008, 23(1):164-168(in Chinese). http://d.wanfangdata.com.cn/Periodical_hkdlxb200801028.aspx [6] 徐启华, 师军, 耿帅.应用快速多分类SVM的航空发动机故障诊断方法[J].推进技术, 2012, 33(6):961-967. http://d.old.wanfangdata.com.cn/Periodical/tjjs201206021XU Q H, SHI J, GENG S.Aero-engine fault diagnosis by a new fast multi-class support vector algorithm[J].Journal of Propulsion Technology, 2012, 33(6):961-967(in Chinese). http://d.old.wanfangdata.com.cn/Periodical/tjjs201206021 [7] 孙伟超, 李文海, 李文峰.融合粗糙集与D-S证据理论的航空装备故障诊断[J].北京航空航天大学学报, 2015, 41(10):1902-1909. http://bhxb.buaa.edu.cn/CN/abstract/abstract13502.shtmlSUN W C, LI W H, LI W F.Avionic devices fault diagnosis based on fusion method of rough set and D-S theory[J].Journal of Beijing University of Aeronautics and Astronautics, 2015, 41(10):1902-1909(in Chinese). http://bhxb.buaa.edu.cn/CN/abstract/abstract13502.shtml [8] 于劲松, 沈琳, 唐荻音.基于贝叶斯网络的故障诊断系统性能评价[J].北京航空航天大学学报, 2016, 42(1):35-40. http://bhxb.buaa.edu.cn/CN/abstract/abstract13670.shtmlYU J S, SHEN L, TANG D Y.Performance evaluation of fault diagnosis system based on Bayesian network[J].Journal of Beijing University of Aeronautics and Astronautics, 2016, 42(1):35-40(in Chinese). http://bhxb.buaa.edu.cn/CN/abstract/abstract13670.shtml [9] TAI A H, CHING W K, CHAN L Y.Detection of machine failure:Hidden Markov model approach[J].Computers & Industrial Engineering, 2009, 57(2):608-619. https://www.sciencedirect.com/science/article/pii/S0360835208002386 [10] GÓMEZ M J, CASTEJÍN C, GARCÍA-PRADA J C.Automatic condition monitoring system for crack detection in rotating machinery[J].Reliability Engineering & System Safety, 2016, 152:239-247. [11] BAJRIC R, ZUBER N, SKRIMPAS G A, et al.Feature extraction using discrete wavelet transform for gear fault diagnosis of wind turbine gearbox[J].Shock & Vibration, 2016, 2016(1):1-10. http://orbit.dtu.dk/en/publications/feature-extraction-using-discrete-wavelet-transform-for-gear-fault-diagnosis-of-wind-turbine-gearbox(c44ca0bc-9202-4ec5-b645-2d2a5ce80cfe).html [12] SIMANI S.Residual generator fuzzy identification for automotive diesel engine fault diagnosis[J].International Journal of Applied Mathematics & Computer Science, 2013, 23(2):419-438. [13] 周利敏, 闫国华, 瞿红春.一种云环境下的发动机健康管理系统及其应用[J].计算机测量与控制, 2014, 22(12):3926-3928. doi: 10.3969/j.issn.1671-4598.2014.12.026ZHOU L M, YAN G H, QU H C.Engine health management system in cloud and its applications[J].Computer Measurement and Control, 2014, 22(12):3926-3928(in Chinese). doi: 10.3969/j.issn.1671-4598.2014.12.026 [14] 单外平. 基于深度信念网络的变速器故障分类识别研究[D]. 广州: 华南理工大学, 2015.SHAN W P. Classification and recognition of transmission fault based on deep belief network[D]. Guangzhou: South China University of Technology, 2015(in Chinese). [15] 刘伟峰, 杨爱兰.基于BIC准则和Gibbs采样的有限混合模型无监督学习算法[J].电子学报, 2011, 39(S1):134-139. http://www.cqvip.com/QK/90131X/2011A03/38153760.htmlLIU W F, YANG A L.Unsupervised learning for finite mixture models based on BIC criterion and Gibbs sampling[J].Acta Electronica Sinica, 2011, 39(S1):134-139(in Chinese). http://www.cqvip.com/QK/90131X/2011A03/38153760.html [16] 李文立, 郭凯红.D-S证据理论合成规则及冲突问题[J].系统工程理论与实践, 2010, 30(8):1422-1432. doi: 10.12011/1000-6788(2010)8-1422LI W L, GUO K H.Combination rules of D-S evidence theory and conflict problem[J].Systems Engineering Theory and Practice, 2010, 30(8):1422-1432(in Chinese). doi: 10.12011/1000-6788(2010)8-1422 [17] 张光明. 航空发动机气路故障诊断技术研究[D]. 沈阳: 沈阳航空工业学院, 2010.ZHANG G M. Research on gas path fault diagnosis for aero-engine[D]. Shenyang: Shenyang Institute of Aeronautical Engineering, 2010(in Chinese). -
下载:
点击查看大图
计量
- 文章访问数: 1520
- HTML全文浏览量: 250
- PDF下载量: 994
- 被引次数: 0
