作者：邓壮壮1, 杨世海2,3, 朱节中1, 李运2,3
作者单位：1. 南京信息工程大学自动化学院, 江苏 南京 210044;
2. 中国科学院国家天文台南京天文光学技术研究所, 江苏 南京 210042;
3. 中国科学院天文光学技术重点实验室(南京天文光学技术研究所), 江苏 南京 210042
关键词：南极望远镜;驱动系统;非预期故障;支持向量机;分类器
摘要：
针对南极望远镜驱动系统的非预期故障检测存在先验信息不足、故障特征难确定和故障样本少等问题，提出一种基于支持向量机(support vector machine,SVM)的非预期故障检测方法。以南极望远镜驱动系统为实验平台故障植入，采集的数据中心化和标准化预处理。基于KNN (K-nearest neighbor)、K-means、BP (back propagation)神经网络和SVM算法建立4种非预期故障检测分类器，将各个算法参数调优，再根据数据特征预测分类。实验结果表明：在相同的实验条件下，基于SVM算法的非预期故障检测分类器性能优于其他3种分类器性能。将该类方法应用于半实物仿真平台，验证该算法可行、有效。
Unanticipated fault diagnosis method for Antarctic telescope drive based on SVM
DENG Zhuangzhuang1, YANG Shihai2,3, ZHU Jiezhong1, LI Yun2,3
1. College of Automation, Nanjing University of Information Science & Technology, Nanjing 210044, China;
2. National Astronomical Observatories/Nanjing Institute of Astronomical Optics & Technology, Chinese Academy of Sciences, Nanjing 210042, China;
3. CAS Key Laboratory of Astronomical Optics & Technology, Nanjing Institute of Astronomical Optics & Technology, Nanjing 210042, China
Abstract: Aiming at the problems of insufficient prior information, difficulty in determining fault characteristics and few fault samples in the unanticipated faults detection of the Antarctic telescope drive system, a new method of unanticipated fault detection based on support vector machine(SVM) is proposed. Taking the drive system of Antarctic telescope as the experimental platform, fault implantation data are collected, the centralized and standardized preprocessing are carried out. Four types of unanticipated fault detection classifiers are established based on KNN(K-nearest neighbor), K-means, BP(back propagation) neural network and SVM algorithm, the parameters of each algorithm are optimized and classification is predicted according to the characteristics of data. The experimental results show that under the same experimental conditions, the performance of the unanticipated fault detection classifier based on SVM algorithm is better than the other three classifiers. The method is applied to the hardware in the loop simulation platform to verify the feasibility and effectiveness of the algorithm.
Keywords: Antarctic telescope;drive system;unanticipated fault;support vector machine;classifier
2023, 49(6):75-81,91 收稿日期: 2021-08-22;收到修改稿日期: 2021-10-24
基金项目: 国家自然科学基金项目(11973065)；天文联合基金重点项目(U1931207)
作者简介: 邓壮壮(1995-),男,江苏泗洪县人,硕士研究生,专业方向为天文仪器的故障诊断、故障隔离等
参考文献
[1] 李恒. 望远镜的多余度跟踪控制系统研究[D]. 北京: 中国科学院大学, 2015.
[2] 李运. 南极望远镜故障诊断及自愈策略研究[D]. 北京: 中国科学院大学, 2018.
[3] 何章鸣. 非预期故障的数据驱动诊断方法研究[D]. 长沙: 国防科学技术大学, 2015.
[4] 胡雷. 面向飞行器健康管理的新异类检测方法研究[D]. 长沙: 国防科学技术大学, 2010.
[5] CORTES C, VAPNIK V. Support-vector networks[J]. Machine Learning, 1995, 20(3): 273-297
[6] PIMENTEL M A F, CLIFTON D A, CLIFTON L, et al. A review of novelty detection[J]. Signal Processing, 2014, 99: 215-249
[7] FERNANDEZ-FRANCOS D, MARTINEZ-REGO D, FONTENLA-ROMERO O, et al. Automatic bearing fault diagnosis based on one-class v-SVM[J]. Computers & Industrial Engineering, 2013, 64(1): 357-365
[8] OLIVIER L E, CRAIG I K. Model-based fault-tolerant control with robustness to unanticipated faults[J]. IFAC-PapersOnLine, 2017, 50(1): 1157-1162
[9] 金星, 邵珠超, 王盛慧. 基于PCA和IGWO-SVM水泥回转窑故障诊断研究[J]. 中国测试, 2017, 43(10): 92-96
[10] 但长林, 李三雁, 张彬. 基于样本熵和SVM的滚动轴承故障诊断方法研究[J]. 中国测试, 2020, 46(11): 37-42
[11] 李运. 极端环境下望远镜驱动系统的非预期状态研究[D]. 北京: 中国科学院大学, 2021.
[12] GUO G, WANG H, BELL D, et al. KNN model-based approach in classification[C] // OTM Confederated International Conferences “On the Move to Meaningful Internet Systems”. Berlin Heidelberg: Springer, 2003.
[13] HAMERLY G, ELKAN C. Learning the k in K-means[J]. Advances in Neural Information Processing Systems, 2004, 16: 281-288
[14] RUMELHART D E, HINTON G E, WILLIAMS R J. Learning representations by back propagating errors[J]. Nature, 1986, 323(6088): 533-536
[15] 李晓燕. 大型天文望远镜驱动控制方法的应用研究[D]. 北京: 中国科学院大学, 2011.
[16] YANG S H, LI Y, WU J J, et al. The unanticipated states of direct drive system of extremely large telescope in extreme environment[C]//Ground-based and Airborne Telescopes VIII. International Society for Optics and Photonics, 2020, 11445: 1144554.
[17] 李运, 杨世海, 吴佳佳, 等. 基于SVM观测器的新异类故障检测方法及应用[J]. 振动、测试与诊断, 2021, 41(2): 292-298
