浙江电力

2018, v.37;No.265(05) 1-6

[打印本页] [关闭]
本期目录(Current Issue) | 过刊浏览(Archive) | 高级检索(Advanced Search)

深度神经网络在配电网公变短期负荷预测中的应用研究
Application of Deep Neural Network in Short-term Load Prediction of Public Transformer of Power Distribution Network

黄宇腾,韩翊,赖尚栋
HUANG Yuteng,HAN Yi,LAI Shangdong

摘要(Abstract):

精准的负荷预测关系着电力系统安全、经济和可靠运行,短期负荷预测一直是电力系统的重要研究方向之一。结合深度学习理论,基于MXNet深度学习框架,采用深度神经网络算法预测配电网公变短期负荷,考虑负荷自身历史运行状态、气象因素、变压器属性、电力用户特征等多重因素影响,对传统电力负荷预测进行了创新和探索,并通过在某省的实际应用效果表明,基于MXNet框架的深度神经网络模型训练效率良好。基于深度神经网络的短期负荷预测模型有很强的泛化能力与通用性,为不同地区、不同类型的公变建立个性化的预测模型提供了可行方法。模型部署于阿里云大数据平台,基于阿里云大数据实现了配电网公变日负荷的实时预测。
Power load prediction accuracy concerns operation safety, economy and reliability, and short-term load forecasting is one of the important research directions of power system. In combination with deep learning theory and based on MXNet deep learning framework, deep neural network algorithm is adopted to predict short-term load of public transformer of distribution networks; in consideration of historic operation status, meteorological factors, transformer features and characteristics of power consumers, the traditional power load prediction is innovated and explored. It is demonstrated through the application in a province that the deep neural network model based on MXNet framework is efficient in training. The short-term load prediction model based on deep neural network is strongly capable of generalization and versatility, providing a feasible method for customized prediction model establishment for different types of public transformers in different areas. The model is deployed in a big data platform of Alibaba Cloud to predict daily load of public transformer based on big data of Alibaba cloud instantly.

关键词(KeyWords): 负荷预测;配电网公变;深度神经网络模型;MXNet;深度学习
load prediction;public transformer of distribution network;deep neural network model;MXNet;deep learning

Abstract:

Keywords:

基金项目(Foundation):

作者(Author): 黄宇腾,韩翊,赖尚栋
HUANG Yuteng,HAN Yi,LAI Shangdong

DOI: 10.19585/j.zjdl.201805001

参考文献(References):

扩展功能
本文信息
服务与反馈
本文关键词相关文章
本文作者相关文章
中国知网
分享