基于领域自适应的卫星工程参数异常检测

doi:10.19678/j.issn.1000-3428.0064512

摘要/Abstract

摘要： 卫星工程参数异常检测是判断卫星与载荷健康状态的重要手段。传统异常检测方法通常针对具体的卫星或载荷设计，在处理新卫星和新载荷时需要重新建模和训练，然而新运行的卫星设备尚未产生足够的历史数据来支撑建模任务。融合深度学习和迁移学习思想，提出一种基于领域自适应的卫星工程参数异常检测方法。利用深度残差收缩网络(DRSN)结构框架，在训练过程中将无标签的目标域卫星工程参数数据加入网络训练过程，有效获取数据的特征表示。对提取到的数据特征使用距离度量，衡量所提取的源域特征与目标域特征之间的差异性，使得两者在特定特征空间上的分布相似，从而增强了DRSN利用已有标签数据对无标签数据的分类能力。在基于空间科学先导专项系列卫星工程参数构建的数据集上的实验结果表明，该方法的准确率达到98.03%，与梯度提升决策树、支持向量机、DRSN方法相比，迁移检测平均准确率分别高出20.95、22.51、15.02个百分点，能够有效对新卫星和新载荷开展快速准确的异常检测。

关键词: 卫星工程参数, 异常检测, 深度学习, 迁移学习, 领域自适应

Abstract: Anomaly detection of satellite engineering parameters is an important means to judge the health status of satellites and payloads. However，conventional anomaly detection methods are trained and designed based on data from a single satellite，achieving poor detection results when applied to new satellites. Additionally，it takes time to gather enough training data from a new satellite in orbit to create a useful model specific to that satellite. To solve this problem，an anomaly detection method based on domain adaptation is proposed. A Deep Residual Shrinkage Network（DRSN） is applied to extract the features of detected data by adding unlabeled detected data into the training process. In addition，a distance measure is used to estimate dissimilarities between source and target domain features.The distance measure can be used to make feature distributions in specific feature spaces similar，thus enhancing DRSN's ability to classify unlabeled data using existing labeled data. Based on engineering datasets from satellites used in the Space Science Pioneer Project，the detection accuracy of the proposed method is 98.03%，which is an improvement of 20.95，22.51，and 15.02 percentage points over the Gradient Boosting Decision Tree（GBDT），Support Vector Machine（SVM），and DRSN，respectively. Thus，this method can provide precise and fast anomaly detection when a new satellite or a new payload is launched.

Key words: satellite engineering parameter, anomaly detection, deep learning, transfer learning, domain adaptation

中图分类号:

TP391.4

王爱玲, 马文臻, 邹自明, 钟佳. 基于领域自适应的卫星工程参数异常检测[J]. 计算机工程, 2023, 49(5): 29-37,47.

WANG Ailing, MA Wenzhen, ZOU Ziming, ZHONG Jia. Anomaly Detection of Satellite Engineering Parameters Based on Domain Adaptation[J]. Computer Engineering, 2023, 49(5): 29-37,47.

http://www.ecice06.com/CN/Y2023/V49/I5/29

图/表 17

20230515184549

20230515184552

20230515184555

20230515184559

20230515184604

20230515184607

20230515184611

20230515184614

20230515184618

20230515184621

20230515184624

20230515184629

20230515184632

20230515184636

20230515184639

20230515184645

20230515184649

参考文献

[1] 王亚坤,杨凯飞,张婕,等.卫星在轨故障案例与人工智能故障诊断[J].中国空间科学技术,2022,42(1):16-29. WANG Y K,YANG K F,ZHANG J,et al.Case study of in-orbit satellite failures and artificial intelligence based failure detection[J].Chinese Space Science and Technology,2022,42(1):16-29.(in Chinese)
[2] 彭喜元,庞景月,彭宇,等.航天器遥测数据异常检测综述[J].仪器仪表学报,2016,37(9):1929-1945. PENG X Y,PANG J Y,PENG Y,et al.Review on anomaly detection of spacecraft telemetry data[J].Chinese Journal of Scientific Instrument,2016,37(9):1929-1945.(in Chinese)
[3] DENNIS D.Automated learning and monitoring of limit functions[EB/OL].[2022-03-14].https://www.researchgate.net/publication/2813766_Automated_Learning_and_Monitoring_of_Limit_Functions.
[4] YAIRI T,KAWAHARA Y,FUJIMAKI R,et al.Telemetry-mining:a machine learning approach to anomaly detection and fault diagnosis for space systems[C]//Proceedings of the 2nd IEEE International Conference on Space Mission Challenges for Information Technology.Washington D.C.,USA.IEEE Press,2006:456-476.
[5] LIU D T,PANG J Y,SONG G,et al.Fragment anomaly detection with prediction and statistical analysis for satellite telemetry[J].IEEE Access,2017,5:19269-19281.
[6] 杨甲森.卫星遥测数据相关性知识发现方法研究[D].北京:中国科学院大学,2019. YANG J S.Research on correlation knowledge discovery method of spacecraft telemetry data[D].Beijing:Chinese Academy of Sciences,2019.(in Chinese)
[7] SHAWE-TAYLOR J,ŽLIČAR B.Novelty detection with one-class support vector machines[M].Berlin,Germany:Springer,2015.
[8] LI Y Q,LEI M J,LIU P P,et al.A novel framework for anomaly detection for satellite momentum wheel based on optimized SVM and Huffman-multi-scale entropy[J].Entropy,2021,23(8):1062.
[9] 董静怡,庞景月,彭宇,等.集成LSTM的航天器遥测数据异常检测方法[J].仪器仪表学报,2019,40(7):22-29. DONG J Y,PANG J Y,PENG Y,et al.Spacecraft telemetry data anomaly detection method based on ensemble LSTM[J].Chinese Journal of Scientific Instrument,2019,40(7):22-29.(in Chinese)
[10] WANG Y,WU Y,YANG Q,et al.Anomaly detection of spacecraft telemetry data using temporal convolution network[C]//Proceedings of IEEE International Instrumentation and Measurement Technology Conference.Washington D.C.,USA:IEEE Press,2021:1-5.
[11] CHEN J F,PI D C,WU Z Y,et al.Imbalanced satellite telemetry data anomaly detection model based on Bayesian LSTM[J].Acta Astronautica,2021,180:232-242.
[12] 王婵,王慧泉,金仲和.皮纳卫星遥测数据异常检测聚类分析方法[J].哈尔滨工业大学学报,2018,50(4):110-116. WANG C,WANG H Q,JIN Z H.Pico-satellite telemetry anomaly detection through clustering[J].Journal of Harbin Institute of Technology,2018,50(4):110-116.(in Chinese)
[13] ZHU Y C,XI D B,SONG B W,et al.Modeling users' behavior sequences with hierarchical explainable network for cross-domain fraud detection[C]//Proceedings of the Web Conference 2020.New York,USA:ACM Press,2020:928-938.
[14] MALLICK T,BALAPRAKASH P,RASK E,et al.Transfer learning with graph neural networks for short-term highway traffic forecasting[C]//Proceedings of the 25th International Conference on Pattern Recognition.Washington D.C.,USA:IEEE Press,2021:10367-10374.
[15] CHEN L.Deep transfer learning for static malware classification[EB/OL].[2022-03-14].https://arxiv.org/abs/1812.07606.
[16] OLIVEIRA J S,SOUZA G B,ROCHA A R,et al.Cross-domain deep face matching for real banking security systems[C]//Proceedings of the 7th International Conference on eDemocracy & eGovernment.Washington D.C.,USA:IEEE Press,2018:21-28.
[17] AHMED U,KHAN A,KHAN S H,et al.Transfer learning and meta classification based deep churn prediction system for telecom industry[EB/OL].[2022-03-14].https://arxiv.org/abs/1901.06091.
[18] 刘飞,陈仁文,邢凯玲,等.基于迁移学习与深度残差网络的滚动轴承快速故障诊断算法[J].振动与冲击,2022,41(3):154-164. LIU F,CHEN R W,XING K L,et al.Fast fault diagnosis algorithm for rolling bearing based on transfer learning and deep residual network[J].Journal of Vibration and Shock,2022,41(3):154-164.(in Chinese)
[19] LI X,ZHENG J,LI M,et al.Frequency-domain fusing convolutional neural network:a unified architecture improving effect of domain adaptation for fault diagnosis[J].Sensors (Basel,Switzerland),2021,21(2):450.
[20] 王晋东,陈益强.迁移学习导论[M].北京:电子工业出版社,2021. WANG J D,CHEN Y Q.Introduction to transfer learning[M].Beijing:Publishing House of Electronics Industry,2021.(in Chinese)
[21] SUN B,SAENKO K.Deep CORAL:correlation alignment for deep domain adaptation[M].Berlin,Germany:Springer,2016.
[22] LECUN Y,BOTTOU L,BENGIO Y,et al.Gradient-based learning applied to document recognition[J].Proceedings of the IEEE,1998,86(11):2278-2324.
[23] ZHAO M H,ZHONG S S,FU X Y,et al.Deep residual shrinkage networks for fault diagnosis[J].IEEE Transactions on Industrial Informatics,2020,16(7):4681-4690.
[24] 马文臻,王爱玲,李旭东,等.基于GBDT的卫星工程参数异常检测[J].计算机系统应用,2022,31(1):138-144. MA W Z,WANG A L,LI X D,et al.Anomaly detection of satellite engineering data based on GBDT[J].Computer Systems & Applications,2022,31(1):138-144.(in Chinese)
[25] LI X D,HU Y,ZHENG J H,et al.Central moment discrepancy based domain adaptation for intelligent bearing fault diagnosis[J].Neurocomputing,2021,429:12-24.

选择文件类型/文献管理软件名称

选择包含的内容