Deep Learning Model for Smart Grid Electricity Price Prediction

MIAO Yu-rong; CHEN Sen-bo

doi:10.13718/j.cnki.xsxb.2021.05.023

Aiming at the problems of accuracy caused by machine learning method's electricity price prediction on large data sets, a deep learning electricity price prediction model based on big data time series for electricity grid electricity price prediction has been proposed in this paper. Firstly, the collected data are preprocessed and normalized, and then feature selection is performed by means of a hybrid module of ReliefF algorithm and mutual information (MI). Secondly, the improved features are given to kernel principal component analysis (KPCA). And lastly, enhanced convolution neural network (ECNN) is used to predict the electricity price effectively. The experimental results show that compared with the existing benchmark scheme, the proposed model can better predict the price of smart grid, and it is helpful for smart grid to better operate and plan power generation.

HTML

在传统电网中大部分电能在发电、输电和配电过程中被浪费，因此引入智能电网来解决该问题. 智能电网就是电网的智能化，它以集成、高速双向通信网络为基础^[1-2]，通过先进的传感和测量技术、设备技术、控制方法以及先进的决策支持系统技术的应用，实现电网可靠、安全、经济、高效、环境友好和使用安全的目标^[3-4]. 智能电网的核心内涵是实现电网的信息化、数字化和自动化^[5]，在电力负荷和电价预测方面的数据科学研究在平衡供需、规划发电高效运行方面发挥了重要作用. 智能电网根据消费者的要求提供响应，从而满足消费者的所有要求，因此会收集大量有关消费者电力消耗的数据^[6].

智能电网的出现催生了众多智能分布式能源的开发，例如电池存储、可调负载和具有双向充电功能的电动汽车^[7-8]. 随着可再生能源资源的发展和分布式能源的引入以及负荷时间序列的不确定性，有必要预测准确的价格，以便进行可靠的有效发电并降低智能电网的电力损耗，从而进行有效的发电调度和用电规划. 然而，在精确的电价预测中，仍然存在计算成本高、可靠性差和执行时间长等问题.

为了解决上述问题，文献[9]提出一种基于深度学习的预测日间用电负荷的模型，该模型首先利用历史用电负荷数据和相关温度参数，通过堆叠去噪自动编码器(Stacked Denoising Auto encoder，SDA)来进行特征提取，然后训练支持向量回归(Support Vector Regression，SVR)模型来预测用电. 但是要训练自动编码器需要大量的数据处理时间和参数超调. 文献[10]提出一种基于经验小波变换、支持向量回归(SVR)和贝叶斯优化的混合电价预测模型，该模型利用小波变换作为处理工具将原始信号分解成特定的模态分量，然后采用SVR和双向长短期记忆作为非线性子序列预测的基本框架，利用贝叶斯来调整参数优化模型性能. 文献[11]提出一种基于混合结构深神经网络的电价预测模型，该模型将卷积神经网络(Convolutional Neural Network，CNN)和长短期记忆(Long and Short Term Memory，LSTM)相结合来预测未来的电价. 模型在CNN包含2个一维卷积层，为了提高训练效率，在第2个卷积层后加入批处理规范化；LSTM分析CNN提取的特征，并估计下一时间点的电价，实现了更好的预测性能.

然而，上述文献存在一定的局限性，需要在避免过度拟合的同时，提高电价预测的性能. 为了克服这些局限性，本文提出一种智能电网中基于大数据时间序列的深度学习电价预测模型. 为了提高分类器的性能，本文使用ReliefF算法和互信息(Mutual Information，MI)进行混合特征选择，降低了特征选择的复杂度和执行时间. 为了避免特征冗余，执行核主成分分析(Kernel Principal Component Analysis，KPCA)提取特征，克服了过度拟合问题. 然后，将有效特征输入到增强卷积神经网络(Enhanced Convolution Neural Network，ECNN)模型进行有效的价格预测.

3. 结语

本文提出一种大数据环境下的时间序列深度学习电价预测模型，解决过度拟合和精度等问题. 实现结果表明，与现有的基准方案相比，本文提出的模型具有更好的性能，有助于智能电网更好地运行和规划发电. 本文的主要贡献是：①降低了特征选择的复杂度和执行时间，减少了计算时间，提高了框架的性能. ②克服了过度拟合的问题，提高了预测模型的精度和性能. ③该框架通过ECNN预测价格，在复杂数据集上表现得更好，而ECNN在大核函数时间序列数据上表现得更好.

未来的工作是将本文模型结合更多的外部特性，增加系统的存储容量，从而进一步提高模型的性能.

Figure (5) Table (3) Reference (11)

Name
	Name cannot be empty!
E-mail
	Mailbox cannot be empty! Mailbox cannot be empty!
Telephone
	Mobile number cannot be empty! Please enter a valid mobile number!
Title

Content
Verification Code

[1]	PENG L, LIU S, LIU R, et al. Effective Long Short-Term Memory with Differential Evolution Algorithm for Electricity Price Prediction [J]. Energy, 2018, 162: 1301-1314. doi: 10.1016/j.energy.2018.05.052 CrossRef Google Scholar
[2]	GHADIMI N, AKBARIMAJD A, SHAYEGHI H, et al. A New Prediction Model Based on Multi-Block Forecast Engine in Smart Grid [J]. Journal of Ambient Intelligence and Humanized Computing, 2018, 9(6): 1873-1888. doi: 10.1007/s12652-017-0648-4 CrossRef Google Scholar
[3]	MUJEEB S, JAVAID N, AKBAR M, et al. Big Data Analytics for Price and Load Forecasting in Smart Grids[C]//Advances on Broadband and Wireless Computing, Communication and Applications. Taichung: Spring Cham, 2019. Google Scholar
[4]	ZAHID M, AHMED F, JAVAID N, et al. Electricity Price and Load Forecasting Using Enhanced Convolutional Neural Network and Enhanced Support Vector Regression in Smart Grids [J]. Electronics, 2019, 8(2): 122-153. doi: 10.3390/electronics8020122 CrossRef Google Scholar
[5]	WANG F, LI K P, ZHOU L D, et al. Daily Pattern Prediction Based Classification Modeling Approach for Day-Ahead Electricity Price Forecasting [J]. International Journal of Electrical Power & Energy Systems, 2019, 105: 529-540. Google Scholar
[6]	AYUB N, JAVAID N, MUJEEB S, et al. Electricity Load Forecasting in Smart Grids Using Support Vector Machine[C]// Advanced Information Networking and Applications. Matsue: Spring Cham, 2020. Google Scholar
[7]	SHAO Z, YANG S L, GAO F, et al. A New Electricity Price Prediction Strategy Using Mutual Information-Based SVM-RFE Classification [J]. Renewable and Sustainable Energy Reviews, 2017, 70: 330-341. doi: 10.1016/j.rser.2016.11.155 CrossRef Google Scholar
[8]	王萍, 李磊, 胡聪, 等. 云计算环境下智能电网短期负荷预测方法[J]. 科学技术与工程, 2018, 18(7): 153-158. doi: 10.3969/j.issn.1671-1815.2018.07.028 CrossRef Google Scholar
[9]	TONG C, LI J, LANG C, et al. An Efficient Deep Model for Day-Ahead Electricity Load Forecasting with Stacked Denoising Auto-Encoders [J]. Journal of Parallel and Distributed Computing, 2018, 117(7): 267-273. Google Scholar
[10]	CHENG H Y, DING X W, ZHOU W N, et al. A Hybrid Electricity Price Forecasting Model with Bayesian Optimization for German Energy Exchange [J]. International Journal of Electrical Power & Energy Systems, 2019, 110(9): 653-666. Google Scholar
[11]	KUO P H, HUANG C J. An Electricity Price Forecasting Model by Hybrid Structured Deep Neural Networks [J]. Sustainability, 2018, 10(4): 1280-1286. doi: 10.3390/su10041280 CrossRef Google Scholar

Message Board

Deep Learning Model for Smart Grid Electricity Price Prediction

Abstract

References

Access History

通讯作者: 陈斌, bchen63@163.com

Article Metrics

Access History

Other Articles By Authors