作者：梁海英, 许昕, 潘宏侠, 付志敏, 张航

作者单位：中北大学机械工程学院, 山西 太原 030051

关键词：供输弹系统;堆栈降噪自动编码器;核主成分分析;信息融合;故障识别

摘要：

对于供输弹系统早期故障信号成分复杂，潜在故障征兆难以识别的问题，提出基于堆叠式降噪自动编码器（SDA）和核主成分分析（KPCA）特征融合的早期故障识别方法。所采集的供输弹系统信号经过去趋势项和五点三次平滑法处理后，首先将不同状态的振动信号和声压信号分别通过SDA进行特征提取；然后用KPCA对提取的振动信号和声压信号特征进行融合；最后运用支持向量机（SVM）对融合前后的特征分别进行识别并对比。试验结果表明，该方法能有效地对供输弹系统早期故障进行识别，且识别准确率达92.4%。

Early fault identification of ammunition supply system based on SDA and KPCA feature fusion
LIANG Haiying, XU Xin, PAN Hongxia, FU Zhimin, ZHANG Hang
Mechanical Engineering Institute, North University of China, Taiyuan 030051, China
Abstract: For the problem that the early fault signal components of the ammunition supply system are complicated and the potential fault signs are difficult to identify, an early fault identification method based on SDA and KPCA feature fusion is proposed. After the collected signal for the ramming system is processed by the detrending term and the five-point three-smoothing method, the vibration signals and the acoustic signals of different states are first extracted by SDA respectively.Then, KPCA is used to fuse the extracted vibration signal and sound pressure signal features. Finally, the SVM is used to classify and compare the features before and after fusion. The test results show that the method can effectively identify the early failure of the ammunition supply system and the recognition accuracy is 92.4%.
Keywords: ammunition supply system;SDA;KPCA;information fusion;fault identification
2019, 45(4):141-145,150  收稿日期: 2018-07-24;收到修改稿日期: 2018-09-01
基金项目: 国家自然科学基金资助项目（51675491）
作者简介: 梁海英(1993-),女,河北张家口市人,硕士研究生,专业方向为信号识别与处理、装备系统检测与诊断
参考文献
[1] 王红玲. 火炮自动供输弹系统故障诊断技术研究[D]. 太原:中北大学, 2010.
[2] 冯旭松, 施伟, 杨雪, 等. 基于堆栈自动编码器的泵站机组故障分析[J]. 人民长江, 2018, 49(8):99-102
[3] 崔江, 唐军祥, 龚春英, 等. 一种基于改进堆栈自动编码器的航空发电机旋转整流器故障特征提取方法[J]. 中国电机工程学报, 2017, 37(19):5696-5706
[4] 潘磊. 基于深度学习网络的风机传动系统主要部件故障诊断的研究[D]. 上海:上海电机学院, 2017.
[5] 许海伦. 基于电流信号分析与小波-支持向量机的火炮自动装填系统故障诊断研究[D]. 太原:中北大学, 2013.
[6] 付志敏, 潘宏侠, 许昕, 等. 基于PCA-KLD的供输弹系统早期故障识别[J]. 机械设计与研究, 2018, 34(2):192-195
[7] 李平. 基于多源异类信息融合的异步电机故障诊断技术研究[D]. 湘潭:湖南科技大学, 2013.
[8] 李学军, 李平, 蒋玲莉, 等. 基于异类信息特征融合的异步电机故障诊断[J]. 仪器仪表学报, 2013, 34(1):227-233
[9] 颜丹. 基于深度自动编码器的特征提取算法研究[D]. 长沙:长沙理工大学, 2016.
[10] 时培明, 梁凯, 赵娜, 等. 基于深度学习特征提取和粒子群支持向量机状态识别的齿轮智能故障诊断[J]. 中国机械工程, 2017, 28(9):1056-1061
[11] ZHU M, GHODSI A. Automatic dimensionality selection from the scree plot via the use of profile likelihood[J]. Computational Statistics & Data Analysis, 2006, 51(2):918-930
[12] 颜丹. 基于深度自动编码器的特征提取算法研究[D]. 长沙:长沙理工大学, 2016.
[13] GHARAHBAGHERI H, IMTIAZ S A, KHAN F. Root cause diagnosis of process fault using KPCA and bayesian network[J]. Industrial & Engineering Chemistry Research, 2017, 56(8):2054-2070
[14] 张颖伟. 基于数据的复杂工业过程监测[M]. 沈阳:东北大学出版社, 2011:13-15.
[15] REN L K, LV W M, JIANG S W, et al. Fault diagnosis using a joint model based on sparse representation and SVM[J]. IEEE Transactions on Instrumentation & Measurement, 2016, 65(10):2313-2320