Power Data Anomaly Detection Based on Holt-winters Model and DBSCAN Clustering

被引：0

作者：

Xiao Y. ^{[1
]}

Zheng K. ^{[1
]}

Yu Z. ^{[2
]}

Zhou M. ^{[1
]}

Li S. ^{[2
]}

Ma Q. ^{[2
]}

机构：

[1] China Southern Power Grid Electric Power Research Institute, Guangzhou, 510080, Guangdong

[2] School of Computer Science and Engineering, South China University of Technology, Guangzhou, 510006, Guangdong

来源：

Dianwang Jishu/Power System Technology | 2020年 / 44卷 / 03期

关键词：

Anomaly detection; DBSCAN clustering; Holt-Winters model; Time series;

D O I：

10.13335/j.1000-3673.pst.2019.0320

中图分类号：

学科分类号：

摘要：

During power system operation, it is inevitable to have abnormal power data generated. How to efficiently detect and identify these abnormal data is a vital component in state estimation of power system and the basis for safety and stability of power system operation. Traditional anomaly detection methods are only for specific power system characteristics, usually with relatively large calculation amount and relatively low accuracy. Aiming at the shortcomings of traditional power data anomaly detection methods, this paper proposes a power data anomaly detection method based on Holt-Winters model and DBSCAN clustering. Holt-Winters model uses historical data to predict current area's electricity consumption, then the residual values are obtained by subtracting the predicted values from real values. Finally, the DBSCAN density clustering algorithm is used to cluster residual items to realize identification of abnormal power data. The anomaly detection method is used to detect the anomaly data of some power system’s areas and compared with three commonly used anomaly data detection models. Results show that Holt-Winters model combined with DBSCAN clustering achieves good results in both detection rate and false positive rate in power anomaly data detection. © 2020, Power System Technology Press. All right reserved.

引用

页码：1099 / 1104

页数：5

共 28 条

[1] Liu H., Dai J., System partition based state estimation algorithm using fast decoupled P-Q and retaining nonlinearity, Power System Technology, 29, 12, pp. 72-76, (2005)
[2] Du Z., Niu Z., Fang W., A block QR based power system state estimation algorithm, Proceedings of the CSEE, 23, 8, pp. 50-55, (2003)
[3] Tang J., A survey of power system state estimation, China New Technologies and Products, 20, pp. 144-145, (2011)
[4] Liu G., Yu E., Xia Z., Estimation and correction of error variance of measurement system, Proceedings of the CSEE, 6, pp. 31-39, (1990)
[5] Wang J., A survey on power system bad data detection and identification, Power & Energy, 6, pp. 813-817, (2015)
[6] Huang Y., Li Y., A new method to detect and identify bad data based on correlativity of measured data in power system, Power System Technology, 30, 2, pp. 70-74, (2006)
[7] Dong Z., Zhao J., Wen F., Et al., From smart grid to energy internet basic concept and research framework, Automation of Electric Power Systems, 38, 15, pp. 1-11, (2014)
[8] Mo Y., Kim H.H., Brancik K., Et al., Cyber-physical security of a smart grid infrastructure, Proceedings of the IEEE, 100, 1, pp. 195-209, (2012)
[9] Pei T., Qi D., Outlier detection method based on compressed time series and Voronoi Diagram for power data, Electric Power Construction, 38, 5, pp. 105-110, (2007)
[10] Ouyang Z., Sun X., Yue D., Hierarchical time series feature extraction for power consumption anomaly detection, Advanced Computational Methods in Energy, Power, Electric Vehicles, and Their Integration, pp. 267-275, (2017)

← 1 2 3 →