计算机工程与应用 ›› 2020, Vol. 56 ›› Issue (9): 240-247.DOI: 10.3778/j.issn.1002-8331.1901-0084

• 工程与应用 • 上一篇    下一篇

电能质量异常数据在线检测方法

刘杰,房俊,雷峰津   

  1. 1.北方工业大学 计算机学院,北京 100144
    2.大规模流数据集成与分析技术北京市重点实验室,北京 100144
  • 出版日期:2020-05-01 发布日期:2020-04-29

On-Line Detection Method for Abnormal Data of Power Quality

LIU Jie, FANG Jun, LEI Fengjin   

  1. 1.Department of Computer, North China University of Technology, Beijing 100144, China
    2.Beijing Key Laboratory on Integration and Analysis of Large-Scale Stream Data, Beijing 100144, China
  • Online:2020-05-01 Published:2020-04-29

摘要:

电网电能质量监测数据大多数监测指标具有周期性变化规律、波动性大等特征,现有的异常数据检测方法在针对此类数据做异常值检测时具有适应性差等问题。针对上述问题,将控制图和时间序列数据预测方法有机结合,提出了基于控制图的动态阈值电能质量异常数据在线检测方法。利用电能质量数据的变化趋势数据替代控制图中心线,将控制图的控制线化直为曲,结合ARIMA模型动态计算控制图的控制线,实现了电能质量异常数据的在线检测。实验结果表明提出的方法是有效的。

关键词: 异常检测, 动态阈值, 控制图, 电能质量

Abstract:

The most indicators data of power quality have the characteristics of periodic variation and wide fluctuation range. The existing abnormal detection methods have the problem of poor adaptability when detecting abnormal values of these data. Aiming at the above problems, this paper organically combines the control chart and time series data prediction method, proposes an online detection method for power quality anomaly data based on the control chart and dynamic threshold. The trend data of power quality data is used to replace the center line of the control chart. The control lines of the control chart are dynamically calculated by combining the control chart with the ARIMA model. Experimental results show that the proposed method is effective.

Key words: abnormal detection, dynamic threshold, control chart, power quality