[1]单新文,李 萌,陶晔波,等.基于深度置信网络的用电量预测方法研究[J].计算机技术与发展,2021,31(增刊):177-182.[doi:10. 3969 / j. issn. 1673-629X. 2021. S. 036]
　SHAN Xin-wen,LI Meng,TAO Ye-bo,et al.Research on Electricity Consumption Forecasting Method Based on Deep Belief Network[J].,2021,31(增刊):177-182.[doi:10. 3969 / j. issn. 1673-629X. 2021. S. 036]
点击复制

基于深度置信网络的用电量预测方法研究()
分享到：

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:
31
期数:
2021年增刊
页码:
177-182
栏目:
应用前沿与综合
出版日期:
2021-12-31

文章信息/Info

Title:
Research on Electricity Consumption Forecasting Method Based on Deep Belief Network
文章编号:
1673-629X(2021)S0177-06
作者:
单新文1 ; 李 萌1 ; 陶晔波1 ; 夏元轶1 ; 许坤洋2 ; 查易艺1
1. 国网江苏省电力有限公司信息通信分公司,江苏 南京 210024;
2. 江苏电力信息技术有限公司,江苏 南京 210024
Author(s):
SHAN Xin-wen1 ; LI Meng1 ; TAO Ye-bo1 ; XIA Yuan-yi1 ; XU Kun-yang2 ; CHA Yi-yi1
1. Information and Communication Company,State Grid Jiangsu Electric Power Co. ,Ltd. ,Nanjing 210024,China;
2. Jiangsu Electric Power Information Technology Co. ,Ltd. ,Nanjing 210024,China
关键词:
用电量预测; 受限玻尔兹曼机; 深度置信网络; 可视化调整
Keywords:
electricity consumption forecasting; restricted Boltzmann machine; deep belief network; visual adjustment
分类号:
TP39;TN99
DOI:
10. 3969 / j. issn. 1673-629X. 2021. S. 036
摘要:
随着智能电网的快速发展和社会总用电量的不断攀升,对用电量进行精准的预测对于保持电网安全稳定运行和提升社会经济效益具有重要意义。 传统方法依赖众多的特征数据,且在预测方法选择上存在诸多缺陷,造成预测精度较差等问题。 文中基于获取到的大量的用电量大数据,建立基于受限玻尔兹曼的深度置信网络模型对用电量数据进行预测,选择平均相对误差以及互相关系数等量化指标对预测模型进行评价,随后提出基于可视化的模型动态训练策略,对深度置信网络预测模型进行动态训练。 基于现场实例对所提方法进行验证,结果表明,基于深度置信网络和可视化调整的用电量预测方法的综合准确率可达 97. 05% ,相对于传统的支持向量机、神经网络以及长短时记忆网络等预测算法具有更高的准确率,对于电力协调分配和电力设备运维检修具有重要意义。
Abstract:
With the rapid development of smart grids and the continuous increase of total electricity consumption in society, accurate forecasting of electricity consumption is of great significance for maintaining the safe and stable operation of the grid and improving social and economic benefits. Traditional methods rely on numerous feature data, and there are many defects in the selection of prediction methods,resulting in problems such as poor prediction accuracy. Based on the large amount of electricity consumption big data obtained,a deep belief network model based on Restricted Boltzmann was established to predict electricity consumption data and select quantitative indicators such as average relative error and cross - correlation coefficient to evaluate the prediction model. At the same time, avisualization-based model dynamic training strategy is proposed to dynamically train the deep belief network prediction model. The proposed method is verified based on field examples,and the results show that the comprehensive accuracy rate of the power consumption forecasting method based on deep belief networks and visual adjustments can reach 97. 05% ,which is compared with traditional support vector machines,neural networks and long and short-term memory networks. The prediction algorithm has higher accuracy rate,which is of great significance for the coordinated distribution of power and the operation and maintenance of power equipment.

相似文献/References:

[1]曾志平[] [],萧海东[],张新鹏[]. 基于DBN的金融时序数据建模与决策[J].计算机技术与发展,2017,27(04):1.
　ZENG Zhi-ping[] [],XIAO Hai-dong[],ZHANG Xin-peng[]. Modeling and Decision-making of Financial Time Series Data with DBN[J].,2017,27(增刊):1.
[2]潘嘉琪,邹俊韬.一种基于深度 RTRBM 的动态网络链路预测方法[J].计算机技术与发展,2020,30(03):1.[doi:10. 3969 / j. issn. 1673-629X. 2020. 03. 001]
　PAN Jia-qi,ZOU Jun-tao.A Dynamic Link Prediction Method Based on Deep Recurrent Temporal Restricted Boltzmann Machine[J].,2020,30(增刊):1.[doi:10. 3969 / j. issn. 1673-629X. 2020. 03. 001]

更新日期/Last Update: 2021-09-10