Completion method for missing time series data of distribution station based on GAKNN method

被引：0
|
作者：
Feng L. [1 ]
Wang S. [1 ]
Liang Q. [1 ]
机构：
[1] School of Mechanical Engineering, Shanghai Jiao Tong University, Shanghai
来源：
Dianli Zidonghua Shebei/Electric Power Automation Equipment | 2021年 / 41卷 / 12期
关键词：
Data completion; Distribution station data; GAKNN; Time series;
D O I：
10.16081/j.epae.202108004
中图分类号：
学科分类号：
摘要：
Data record loss may occur in the process of time series data acquisition, transmission and storage in distribution station, which affects high-level data analysis and processing to a certain extent. To solve this problem, a completion method for missing data based on GAKNN(Gray Adaptive K-Nearest Neighbor) method is proposed. Firstly, the time series features are constructed. Then, the nearest neighbor points are selected by the threshold based on simple KNN(K-Nearest Neighbor) method, and the weight coefficients of the nearest neighbor points are calculated combining with the gray correlation coefficient. Finally, the missing data can be completed in turn. With the electric power data sample of a city in Jiangsu Province, the test results show that the missing data completion results of GAKNN method are better than those of other methods, and the completed samples have lower errors in deep learning prediction. © 2021, Electric Power Automation Equipment Press. All right reserved.
引用
收藏
页码：187 / 192
页数：5
相关论文
共 15 条
  • [1] SUN Guoqiang, SHEN Peifeng, ZHAO Yang, Et al., Intelligent recognition of power grid monitoring alarm event combining knowledge base and deep learning, Electric Power Automation Equipment, 40, 4, pp. 40-47, (2020)
  • [2] CHENG Wanwei, Study on the interpolation method of time series missing value, (2018)
  • [3] SUN Jian, LIAO Haitao, UPADHYAYA B R., A robust functional-data-analysis method for data recovery in multi-channel sensor systems, IEEE Transactions on Cybernetics, 44, 8, pp. 1420-1431, (2014)
  • [4] FARHANGFAR A, KURGAN L, DY J., Impact of imputation of missing values on classification error for discrete data, Pattern Recognition, 41, 12, pp. 3692-3705, (2008)
  • [5] RUBIN D B., Multiple imputation after 18+ years, Journal of the American Statistical Association, 91, 434, pp. 473-489, (1996)
  • [6] CAMPION W M, RUBIN D B., Multiple imputation for nonres-ponse in surveys, Journal of Marketing Research, 26, 4, (1989)
  • [7] ZHANG Ruoyu, QI Bo, ZHANG Peng, Et al., Method for interpolating monitoring data of dissolved gas in oil for power transformer state assessment, Electric Power Automation Equipment, 39, 11, pp. 181-187, (2019)
  • [8] ASHRAPHIJUO M, AGGARWAL V, WANG X D., A characteri-zation of sampling patterns for low-tucker-rank tensor comple-tion problem, 2017 IEEE International Symposium on Infor-mation Theory(ISIT), pp. 531-535, (2017)
  • [9] QIAO Wenyu, LI Ye, LIU Haoyu, Et al., Missing load data com-pletion based on curve similarity and low-rank matrix, Electric Power Construction, 41, 1, pp. 32-38, (2020)
  • [10] YIN Ling, YIN Jingyuan, SUN Xiankun, Et al., Reconstruction of gappy GPS coordinate time series based on long short-term memory networks, Journal of Geomatics Science and Technology, 35, 4, pp. 331-336, (2018)
12