Identification of the Bad Data in Power System Based on Deep Learning Neural Network

PAN Zhimin; YU Shuiqiong; LIANG Yunhua

English Version Reader LoginJournal Login

您当前的位置：

首页 >

文章列表页 >

Identification of the Bad Data in Power System Based on Deep Learning Neural Network

更新时间：2025-11-18

- Identification of the Bad Data in Power System Based on Deep Learning Neural Network
- Vol. 38, Issue 6, Pages: 16-19(2018)
- 作者机构：
  
  1. 国网湖南省电力有限公司
  2. 国网湖南省电力有限公司检修公司
- 作者简介：
- 基金信息：
- DOI：
  CLC： TM711
- Published：2018
- 稿件说明：
移动端阅览
[1]潘志敏,俞水琼,梁运华.基于深度学习网络的电力系统不良数据辨识算法研究[J].湖南电力,2018,38(06):16-19.

PAN Zhimin, YU Shuiqiong, LIANG Yunhua. Identification of the Bad Data in Power System Based on Deep Learning Neural Network[J]. 2018, 38(6): 16-19.
[1]潘志敏,俞水琼,梁运华.基于深度学习网络的电力系统不良数据辨识算法研究[J].湖南电力,2018,38(06):16-19. DOI：

PAN Zhimin, YU Shuiqiong, LIANG Yunhua. Identification of the Bad Data in Power System Based on Deep Learning Neural Network[J]. 2018, 38(6): 16-19. DOI：

摘要

为提高电力系统的稳定性和安全性

本文提出一种基于深度学习网络的电力系统不良数据辨识算法。为消除不同量纲数据所引起的差异

算法首先将采集的电力数据进行归整

然后构建多层深度循环学习网络进行不良数据辨识。学习网络分输入层、隐藏层和输出层三层

其中隐藏层包含128个神经元

输出层采用softmax分类器。针对不同神经网络的仿真实验显示

相比于BP神经网络和自组织模糊神经网络

深度循环学习网络的不良数据检测效率最高。针对实测数据的实验显示

相比传统的标准残差法和量测量突变法

不良数据检测率分别提高17.9%和6.4%

辨识率提高17.2%和8.9%。

Abstract

In order to improve the stability and security of power system

an identification algorithm of the bad data in power system based on deep learning neural network is proposed.In order to eliminate the different dimensions of data

the power data is firstly normalized.Then

the multi-layer deep learning neural network is constructed for bad data identification.The network is divided into three layers:the input layer

the hidden layer and the output layer.The hidden layer contains 128 neurons

and the output layer adopts the softmax classifier.Simulation experiments for different neural networks show that the detection efficiency of bad data for deep learning neural network is the highest compared with BP neural networks and self-organizing fuzzy neural networks.The experimental results for real data show that compared with the traditional standard residual method and the measurement method

the detection rate of bad data is improved by 17.9% and 6.4%

and the recognition rate is increased by 17.2% and 8.9% respectively.

关键词

Keywords

references

Views

185

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Implementation of XML-based dynamic web page to release information in power system

Application and research on state estimation based on instantaneous reactive power theory

A practical improvement of complex fault calculation based on canonical algorithm in power system

Application of fault traveling wave positioning system in Zhuzhou power grid

Fuzzy algorithm of bi-objective daily active power economic load distribution

Related Author

LI Xin-ran

LIU You-qiang

YANG Feng

DONG Zhao-xia

WU Lin-jun

YANG Feng

DONG Zhao-xia

HAN Ao-yang

Related Institution

湖南大学电气与信息工程学院湖南长沙410082

Electrical and Information Engineering Institution, Hunan University

华中科技大学水电仿真中心湖北武汉430074

Hydropower & Information Engineering College, HUST

华中科技大学水电数字化工程与仿真中心湖北武汉430074

AI问答

⁰