School of Economics and Management, North China Electric Power University, Beijing 102206, China
Show less
文章历史+
Received
Published
Revised
2024-12-16
2028-07-28
2025-10-15
Online Date
2026-09-20
PDF (9401K)
摘要
日度电费收入预测对于电费管理以及电力保供具有重要价值。为增强电费预测所用样本的时效性,量化预测结果的不确定性,提出一种基于滚动时间窗的自适应混合区间预测模型。首先,设置滚动时间窗长度,并应用所提改进变分模态分解-模糊熵(improved variational mode decomposition-fuzzy entropy,IVMD-FE)的组合数据预处理策略对电费数据进行分解与重构,实现对电费数据特征高效且准确的提取。在此基础上,利用模糊信息粒子提取出各重构序列的最大值与最小值以进行区间构造。最后,应用鲸鱼优化回声状态网络(whale optimization algorithm-echo state network,WOA-ESN)模型预测不规则序列,差分自回归移动平均(auto regressive integrated moving average ARIMA)模型预测常规序列,通过更新滚动时间窗的数据,动态获取电费区间。研究结果表明,所提模型获取的电费预测区间具有优良的覆盖率与区间宽度,能够准确覆盖实际电费,为电费回收提供更多有价值的决策信息。
Abstract
Daily electricity revenue forecasting is of great value for electricity tariff management as well as electricity supply assurance. To enhance the timeliness of the samples used for electricity tariff forecasting and to quantify the uncertainty of the forecasting results, an adaptive hybrid interval forecasting model based on a rolling time window was proposed. Firstly, the rolling time window length was set, and the proposed combined data preprocessing strategy of improved variational mode decomposition-fuzzy entropy (IVMD-FE) was applied to decompose and reconstruct the tariff data, so as to achieve efficient and accurate extraction of tariff data features. On this basis, the maximum and minimum values of each reconstructed sequence was employed using fuzzy information particles for interval construction. Finally, the optimization algorithm echo state network (WOA-ESN) model was applied to predict irregular sequences, and the autoregressive integrated moving average model (ARIMA) was used to predict regular sequences, and the tariff intervals were dynamically obtained by updating the data in the rolling time window.The results show that the proposed model obtains electricity rate prediction intervals with excellent coverage and interval width, which can accurately cover the actual electricity revenue and provide more valuable decision-making information for electricity rate recovery.
ZHANG Jin-liang, LIU Zi-yi, WANG Zhao-yang, LÜ Ya-nan.
Adaptive Daily Electricity Revenue Interval Forecasting Model Based on Sliding Time Window[J].
Science Technology and Engineering, 2026, 26(3): 1115-1127 DOI:10.12404/j.issn.1671-1815.2409332
传统的人工神经网络具有很强的非线性映射能力,但存在局部最优、适应性差等缺陷。为更好的捕捉日度电费数据较强的非线性特性,采用回声状态网络(echo state network,ESN) [30]对非线性序列进行预测,ESN突出的优势是采用大规模随机连接的网络替代传统递归神经网络中的隐藏层,增强了模型的适应能力,避免陷入局部最优[31]。运行原理如下:
National Energy Administration. 2024 electricity consumption of the whole society[EB/OL]. (2025-01-20)[2025-04-06] https://www.nea.gov.cn/20250120/4f7f249bac714e7693adecac996d742f/c.html.
MaLulu. Research on electricity tariff mechanisms adapted to new power system[J]. Price: Theory & Practice, 2024(12): 196-202, 232.
[5]
ZhangJ, LiuZ, ChenT. Interval prediction of ultra-short-term photovoltaic power based on a hybrid model[J]. Electric Power Systems Research2023, 216: 109035.
LiLongfei, ZhongHaiwang, ZhangGuanglun, et al. An industry monthly electricity forecasting method based on causal analysis[J]. Power Grid Technology, 2023, 47(9): 3629-3638.
ChenYucong, BaiXiaoqing. Forecasting model of day-ahead market electricity price considering 2D variation of time series[J]. Proceedings of the CSU-EPSA, 2024, 36(7): 22-29.
LiuLiyan, ZouXiaoyan. Comparative study of GARCH family models in electricity market tariff forecasting[J]. Power System Protection and Control, 2016, 44(4): 57-63.
[12]
TanZ, ZhangJ, WangJ, et al. Day-ahead electricity price forecasting using wavelet transform combined with ARIMA and GARCH models[J]. Applied Energy2010, 87: 3606-3610.
ZhaoYaxue, WangXu, JiangChuanwen, et al. A short-term electricity price forecasting method based on maximum information coefficient correlation analysis and improved multilevel gated LSTM[J]. Chinese Journal of Electrical Engineering, 2021, 41(1): 135-146, 404.
LiChangzu, NiuDongxiao, ZhangXinyan, et al. An improved particle swarm optimization BP neural network-based model for predicting electrical energy substitution potential in Zhejiang[J]. Science Technology and Engineering, 2020, 20(13): 5173-5179.
[17]
WangK, YuM, NiuD, et al. Short-term electricity price forecasting based on similarity day screening, two-layer decomposition technique and Bi-LSTM neural network[J]. Applied Soft Computing2023, 136: 110018.
[18]
HeidarpanahM, HooshyariporF, FazeliM. Daily electricity price forecasting using artificial intelligence models in the Iranian electricity market[J]. Energy, 2023, 263: 126011.
[19]
ZhangJ, TanZ, WeiY. An adaptive hybrid model for short term electricity price forecasting[J]. Applied Energy2020, 258: 114087.
HuangGuoquan, YanYuting, LiHui, et al. Modeling short-term daily power forecasting for industry customers based on dual decomposition[J]. Southern Power Grid Technology, 2022, 16(11): 37-45.
[22]
孙越. 基于RF-XGBoost模型的电费收入预测研究[D]. 延吉: 延边大学, 2022.
[23]
SunYue. Research on electricity revenue forecasting based on RF-XGBoost model[D]. Yanji: Yanbian University, 2022.
YinHao, DingWeifeng, ChenShun, et al. Long and short term memory network-based longitudinal crossover algorithm for predicting electricity prices before market days with high proportion of new energy sources[J]. Power Grid Technology, 2022, 46(2): 472-480.
LiuLi, WangYanbo, PangXinfu, et al. A comprehensive forecasting method for monthly electricity sales based on STL model[J]. Control Engineering, 2020, 27(11): 1930-1936.
ZhangKunming, CaiShanshan, ZhangTianhan, et al. Medium- and long-term industry load forecasting method consideringmulti-dimensional temporal features[J]. Automation of Electric Power Systems, 2023, 47(20): 104-114.
WangWei, FengBin, HuangGang, et al. Short-term net load forecasting based on self-attention encoder and deep neural network[J]. Proceedings of the CSEE, 2023, 43(23): 9072-9084.
LiMiaotong, WangYong, XiaoLing, et al. Explainable deep learning based power load forecasting model[J]. Foreign Electronic Measurement Technology, 2023, 42(4): 66-73.
WangHongtao, ZouBin. Electricity price interval prediction based on dynamic Bayesian networks[J]. Power System Protection and Control, 2022, 50(5): 117-127.
ZhaoHuiru, ZhangShiying, ZhaoYihang, et al. Short-term power load interval forecasting based on adaptive noise-complete empirical modal decomposition-sample entropy-long-term memory neural network and kernel density estimation[J]. Modern Electricity, 2021, 38(2): 138-146.
TangZhenpeng, WuJunchuan, ZhangTingting, et al. Research on grain futures price forecasting based on quadratic decomposition and integrated learning[J]. Systems Engineering-Theory & Practice, 2021, 41(11): 2837-2849.
[44]
ZhangJ, WangS Y, TanZ F, et al. An improved hybrid model for short term power load prediction[J]. Energy, 2023, 268: 126561.
[45]
ZhangJ, TanZ, WeiY. An adaptive hybrid model for short term electricity price forecasting[J]. Applied Energy, 2020, 258: 114087.
ZhangJinliang, LiuZiyi. Ultra-short-term wind speed interval prediction based on hybrid model[J]. Power System Protection and Control, 2022, 50(22): 49-58.
WuMenglong, YeYicheng, HuNanyan, et al. Fuzzy information granulation-based interval prediction of mining safety production situation[J]. Chinese Journal of Safety Science, 2021, 31(9): 119-127.
XiongJinghua, RuJing. Research on the combined model of exchange rate forecasting based on random forest algorithm and fuzzy information granulation[J]. Quantitative Economic and Technical Economics Research, 2021, 38(1): 135-156.
[52]
JaegerH, HaasH. Harnessing nonlinearity: predicting chaotic systems and saving energy in wireless communication[J]. Science, 2004, 304(5667): 78-80.
[53]
GhimireS, Nguyen-HuyT, AL-MusaylhM S, et al. A novel approach based on integration of convolutional neural networks and echo state network for daily electricity demand prediction[J]. Energy, 2023, 275: 127430.
LiJiahao, WangQingyu, FanYuelin, et al. Whale optimization algorithm-bidirectional long and short-term memory neural network for the prediction of mechanical remaining life of circuit breakers[J]. High Voltage Technology, 2024, 50(1): 250-262.
[56]
ZhangJ, TanZ, WeiY. An adaptive hybrid model for day-ahead photovoltaic output power prediction[J]. Journal of Cleaner Production, 2020;244: 118858.
FanLiwei, DongHuanhuan, JianLing. A study on integrated prediction of carbon market price decomposition based on rolling time window[J]. China Management Science, 2023, 31(1): 277-286.