Science & Technology Review >

2024 , Vol. 42 >Issue 16: 114 - 125

DOI: https://doi.org/10.3981/j.issn.1000-7857.2022.10.01513

Papers

Reactor coolant pump status anomaly detection method based on multi-scale fully convolutional networks

  • GONG An ,
  • WEI Jinming
Expand
  • School of Qingdao Software, School of Computer Science and Technology, China University of Petroleum(East China), Qingdao 266580, China

Received date: 2022-10-09

  Revised date: 2022-11-11

  Online published: 2024-09-05

Fold

Abstract

The large number of types of reactor coolant pump condition sensors leads to three difficulties in current main pump condition anomaly detection: difficulty in modeling complex condition data, difficulty in detecting early abnormalities, and difficulty in assessing the degree of abnormality. The study of rapidly developing deep learning techniques such as neural networks can provide new ideas to solve these problems. To this end, a codec network structure anomaly detection method MSFCNAD (multi-scale FCN-based anomaly detection) based on multi-scale fully convolutional neural networks is proposed. Based on the multivariate temporal characteristics of the main pump state, this method uses the full convolutional neural network codec for pixel-level training to precisely locate the abnormal range of reactor coolant pump state data. At the same time, taking into account the main pump state abnormal time characteristics, the multi-scale feature matrix of reactor coolant pump state is extracted, and the abnormal extent is judged by the graded abnormal range detected by the feature matrix of different scales. On this basis, experiments are conducted using real nuclear main pump data to compare the classification effects of several models such as ARMA, BiLSTM, FCN and AEs. The results show that the MSFCNAD model outperforms the models listed in the paper in terms of recall and F1 score, which are 78.44% and 80.30%, respectively, better than the highest 77.53% and 69.74% of the other models. The experimental results show that this method has better performance compared with other anomaly detection methods, and it can also judge the severity of anomalies by the degree of anomalies and prioritize maintenance processing.

Key words： anomaly detection; multivariate time series; fully convolutional neural networks; multiscale analysis; deep learning

Cite this article

GONG An , WEI Jinming . Reactor coolant pump status anomaly detection method based on multi-scale fully convolutional networks[J]. Science & Technology Review, 2024 , 42(16) : 114 -125 . DOI: 10.3981/j.issn.1000-7857.2022.10.01513

References

[1] Maurya C K, Toshniwal D. Anomaly detection in nuclear power plant data using support vector data description[C]//Proceedings of the 2014 IEEE Students'Technology Symposium. Piscataway:IEEE Press, 2014:82-86.
[2] Calivá F, De Ribeiro F S, Mylonakis A, et al. A deep learning approach to anomaly detection in nuclear reac-tors[C]//2018 International Joint Conference on Neural Networks (IJCNN). Piscataway:IEEE Press, 2018:1-8.
[3] You D D, Shen X C, Liu G J, et al. Signal anomaly identi-fication strategy based on Bayesian inference for nuclear power machinery[J]. Mechanical Systems and Signal Pro-cessing, 2021, 161:107967.
[4] Cheng M, Li Q, Lv J M, et al. Multi-scale LSTM model for BGP anomaly classification[J]. IEEE Transactions on Services Computing, 2021, 14(3):765-778.
[5] Sabokrou M, Fayyaz M, Fathy M, et al. Deep-anomaly:fully convolutional neural network for fast anomaly detec-tion in crowded scenes[J]. Computer Vision and Image Understanding, 2018, 172:88-97.
[6] 陈磊,秦凯,郝矿荣.基于集成LSTM-AE的时间序列异常检测方法[J].华中科技大学学报(自然科学版), 2021, 49(11):35-40.
[7] Szczotka A B, Shakir D I, Clarkson M J, et al. Zero-shot super-resolution with a physically-motivated downsam-pling kernel for endomicroscopy[J]. IEEE Transactions on Medical Imaging, 2021, 40(7):1863-1874.
[8] Shelhamer E, Long J, Darrell T. Fully convolutional net-works for semantic segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(4):640-651.
[9] Jian J, Xia W, Zhang R, et al. Multiple instance convolu-tional neural network with modality-based attention and contextual multi-instance learning pooling layer for effec-tive differentiation between borderline and malignant epi-thelial ovarian tumors[J]. Artificial Intelligence in Medi-cine, 2021, 121:102194.
[10] Labonne M, Olivereau A, Polve B, et al. Unsupervised protocol-based intrusion detection for real-world net-works[C]//2020 International Conference on Computing, Networking and Communications (ICNC). Piscataway:IEEE Press, 2020:299-303
[11] Mao W D, Lin J, Wang Z F. F-DNA:Fast convolution architecture for deconvolutional network acceleration[J]. IEEE Transactions on Very Large Scale Integration (VL-SI) Systems, 2020, 28(8):1867-1880.
[12] Ergen T, Kozat S S. Unsupervised anomaly detection with LSTM neural networks[J]. IEEE Transactions on Neural Networks and Learning Systems, 2020, 31(8):3127-3141.
[13] Malhotra P, Ramakrishnan A, Anand G, et al. LSTMbased encoder-decoder for multi-sensor anomaly detec-tion[DB/OL]. arXiv perprint:1607.00148, 2016.
[14] Tejasri M, Lakshmi K S, Narayan K G R, et al. Unsuper-vised distance-based anomaly disclosure in RNN[J]. In-ternational Journal of Computer Sciences and Engineer-ing, 2018, 6(3):439-441.
[15] 田园,马文.基于Attention-BiLSTM的电网设备故障文本分类[J].计算机应用, 2020, 40(增刊2):24-29.
[16] 侯修群,蒋庆磊,包彬彬,等.基于相关系数的核电主泵振动异常定位方法研究[J].核科学与工程, 2021, 41(5):920-928.
[17] 周勇,朱鹏树,陈星,等.核电厂应急柴油发电机组监测数据异常检测综述[J].仪器仪表用户, 2020, 27(10):52-55, 78.
[18] 向玲,朱浩伟,丁显,等.基于CAE与BiLSTM结合的风电机组齿轮箱故障预警方法研究[J].动力工程学报, 2022, 42(06):514-521.
[19] Galante L, Banisch R. A comparative evaluation of anomaly detection techniques on multivariate time series data[J]. International Journal of Computational Science and Engineering, 2019, 18:17-29.
[20] 夏英,韩星雨.融合统计方法和双向卷积LSTM的多维时序数据异常检测[J].计算机应用研究, 2022, 39(5):1362-1367, 1409.
[21] 于旭,何亚东,杜军威,等.一种结合显式特征和隐式特征的开发者混合推荐算法[J].软件学报, 2022, 33(5):1635-1651.
[22] Hou J, Ni K, Hawari A. An artificial neural network based anomaly detection algorithm for nuclear power plants[J]. Transactions, 2019, 120(1):219-222.
[23] Bhattacharya G, Ghosh K, Chowdhury A S. Granger cau-sality driven AHP for feature weighted kNN[J]. Pattern Recognition, 2017, 66:425-436.
[24] Abdulaal A, Liu Z H, Lancewicki T. Practical approach to asynchronous multivariate time series anomaly detec-tion and localization[C]//Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery&Data Mining. New York:ACM, 2021:2485-2494.
[25] Zhang C X, Song D J, Chen Y C, et al. A deep neural network for unsupervised anomaly detection and diagno-sis in multivariate time series data[J]. Proceedings of the AAAI Conference on Artificial Intelligence, 2019, 33(1):1409-1416.
[26] Jin M, Liu Y, Zheng Y, et al. Anemone:Graph anoma-ly detection with multi-scale contrastive learning[C]//Proceedings of the 30th ACM International Conference on Information&Knowledge Management. New York:ACM, 2021:3122-3126.
[27] Cai R C, Xu B Y, Yang X Y, et al. An encoder-decoder framework translating natural language to database que-ries[DB/OL]. arXiv perprint:1711.06061, 2017.
[28] Chen Y R, Zhang H, Wang Y N, et al. MAMA net:multi-scale attention memory autoencoder network for anomaly detection[J]. IEEE Transactions on Medical Im-aging, 2021, 40(3):1032-1041.
[29] Shalyga D, Filonov P, Lavrentyev A. Anomaly detection for water treatment system based on neural network with automatic architecture optimization[DB/OL]. arXiv pre-print arXiv:1807.07282, 2018.
[30] Kumar A, Narapareddy V T, Aditya Srikanth V, et al. Sarcasm detection using multi-head attention based bidi-rectional LSTM[J]. IEEE Access, 2020, 8:6388-6397.
[31] Kadri F, Harrou F, Chaabane S, et al. Seasonal ARMAbased SPC charts for anomaly detection:Application to emergency department systems[J]. Neurocomputing, 2016, 173:2102-2114.
[32] Chen X, Feng F, Wu J, et al. Anomaly detection for drinking water quality via deep biLSTM ensemble[C]//Proceedings of the Genetic And Evolutionary Computa-tion Conference Companion. New York:ACM, 2018:3-4.
[33] Ye H, Ma X P, Pan Q F, et al. An adaptive approach for anomaly detector selection and fine-tuning in time series[DB/OL]. arXiv perprint:1907.07843, 2019.
[34] Xiao Q F, Shao S K, Wang J. Memory-augmented adver-sarial autoencoders for multivariate time-series anomaly detection with deep reconstruction and prediction[DB/OL]. arXiv perprint:2110.08306, 2021.
[35] Liu Y X, Lin Y F, Xiao Q F, et al. Self-adversarial vari-ational autoencoder with spectral residual for time series anomaly detection[J]. Neurocomputing, 2021, 458:349-363.
[36] Han C, Rundo L, Murao K, et al. MADGAN:Unsuper-vised medical anomaly detection GAN using multiple ad-jacent brain MRI slice reconstruction[J]. BMC Bioinfor-matics, 2021, 22(2):1-20.
[37] Kao I F, Zhou Y L, Chang L C, et al. Exploring a Long Short-Term Memory based Encoder-Decoder framework for multi-step-ahead flood forecasting[J]. Journal of Hy-drology, 2020, 583:124631.
Options
Outlines

/

〈 〉

Contact

  • Add：No. 86 Xueyuan South Road, Haidian District, Beijing　 Postal Code：100081
  • Tel: +86-10-62138113　Fax: +86-10-62138113
  • E-mail：kjdbbjb@cast.org.cn

    • Visited

      Total visitors:
      Visitors of today:
      Now online:
    Copyright © Science & Technology Review, All Rights Reserved.