大规模用电数据流的快速聚类和异常检测技术

WANG Guilan

您当前的位置：

首页 >

文章列表页 >

大规模用电数据流的快速聚类和异常检测技术

更新时间：2025-08-13

- 大规模用电数据流的快速聚类和异常检测技术
- Automation of Electric Power Systems Vol. 40, Issue 24, Pages: 27-33(2016)
- 作者机构：
  
  华北电力大学电气与电子工程学院,河北省,保定市,071003
- 作者简介：
- 基金信息：
- DOI：
  CLC：
- Published：2016
- 稿件说明：
移动端阅览
WANG Guilan. [J]. Automation of electric power systems, 2016, 40(24): 27-33.
DOI：

WANG Guilan. [J]. Automation of electric power systems, 2016, 40(24): 27-33. DOI：

摘要

对近年来在电力系统中出现的大规模数据流进行了探讨

目的是利用流式计算技术提高系统的实时性和安全性。针对大规模用电信息采集中用电数据流的快速聚类和异常检测技术展开研究。结合分布式流式计算平台Spark Streaming

基于用电行为在纵向时间和横向空间上表现出的聚类特性

即同类用户具有相似用电模式和同一用户历史数据具有相似性

设计并实现了流式DBSCAN聚类算法

以实现对大规模用电数据流的快速异常检测。设计并搭建了支持大规模数据流处理的实验环境

证明了算法的有效性。

Abstract

With the large-scale data stream recently emerging in power systems

utilizing stream computing technology to improve power system real-time and safety has become a critical requirement. For the large-scale data stream of power consumption information collection system

fast clustering technology and anomaly detection technology for streaming data is studied. With reference to the distributed stream computing platform Spark Streaming

Streaming DBSCAN algorithm is designed and implemented by taking advantage of longitudinal time and transverse space clustering features exhibited in the electricity consumption behavior

which means that the same cluster of users have similar power consuming pattern

and one user has similar historical power consuming data. The streaming DBSCAN algorithm is able to achieve fast anomaly detection of a large-scale power data stream. The experimental environment in support of large-scale data stream processing is set up

which can support and validate the effectiveness of the algorithm. This work is supported by National Natural Science Foundation of China(No. 51277074)

Hebei Provincial Natural Science Foundation of China(No. F2014502069)and the Fundamental Research Funds for the Central Universities(No. 13MS103).

关键词

Keywords

references

Views

2854

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Portrait-based Assessment on Demand Response Potential of Industrial Parks

Deep Learning Based Anomaly Detection for Application-layer Message of Power Industrial Control Communication Traffic

Identification Method of Electricity Theft Based on Long Short-term Memory Autoencoder with Load Peak Features

Frequency Anomaly Detection and Type Identification of PMU Based on Improved Dynamic Time Warping

Related Author

JU Ping

HUANG Qifeng

JIANG Tingyu

FAN Yuhui

JIANG Jinliang

DU Pengcheng

WANG Wenbo

LIU Xuan

Related Institution

Marketing Service Center of State Grid Jiangsu Electric Power Co., Ltd.

College of Energy and Electrical Engineering, Hohai University

Beijing Smartchip Microelectronics Technology Company

School of Electrical and Information Engineering, Hunan University

State Grid Hunan Electric Power Company

⁰