基于自注意力特征提取的光伏功率组合概率预测

王家乐; 张耀; 林帆; 周一丹; 孙乾皓

doi:10.19912/j.0254-0096.tynxb.2023-1286

PDF(2072 KB)

太阳能学报 ›› 2024, Vol. 45 ›› Issue (12) : 123-131. DOI: 10.19912/j.0254-0096.tynxb.2023-1286

基于自注意力特征提取的光伏功率组合概率预测

王家乐, 张耀, 林帆, 周一丹, 孙乾皓

作者信息 +

COMBINING PROBABILISTIC PREDICTION OF PV POWER BASED ON SELF-ATTENTION FEATURE EXTRACTION MECHANISM

Wang Jiale, Zhang Yao, Lin Fan, Zhou Yidan, Sun Qianhao

Author information +

文章历史 +

摘要

为准确预测光伏出力的概率分布,提出一种基于自注意力机制特征提取的光伏功率组合概率预测方法,可以弥补传统线性组合预测方法在信息利用率及灵活性等方面的缺陷。首先构建兼具同质模型与异质模型的概率预测模型池,之后通过基于自注意力机制的特征提取模块对基模型预测结果进行自适应特征提取,最后将外部特征与所提取基模型特征一并输入残差全连接网络中,以预测分位数增量的形式实现单调分位数预测。基于公开光伏数据集进行算例分析,相关算例结果表明,相较于个体模型及传统线性组合方法,所提方法在概率预测方面具有更好的综合表现。

Abstract

To accurately predict the probability distribution of photovoltaic output, this paper proposes a novel approach for probabilistic forecasting of photovoltaic power based on self-attention feature extraction mechanism. This method addresses the limitations of traditional linear combining prediction methods in terms of information utilization and flexibility. Firstly, a model pool is constructed which includes both homogeneous models and heterogeneous models. Subsequently, a feature extraction module based on self-attention mechanism is employed to adaptively extract features from the predictions of the base models. Lastly, the extracted features from the base models and the external features are input into a residual fully-connected network to achieve monotonic quantile prediction through quantile increment prediction. Experimental analysis using publicly available photovoltaic datasets demonstrates that the proposed method outperforms individual models and traditional linear combination methods in terms of overall performance.

导出引用

王家乐, 张耀, 林帆, 周一丹, 孙乾皓. 基于自注意力特征提取的光伏功率组合概率预测[J]. 太阳能学报. 2024, 45(12): 123-131 https://doi.org/10.19912/j.0254-0096.tynxb.2023-1286

Wang Jiale, Zhang Yao, Lin Fan, Zhou Yidan, Sun Qianhao. COMBINING PROBABILISTIC PREDICTION OF PV POWER BASED ON SELF-ATTENTION FEATURE EXTRACTION MECHANISM[J]. Acta Energiae Solaris Sinica. 2024, 45(12): 123-131 https://doi.org/10.19912/j.0254-0096.tynxb.2023-1286

中图分类号： TM615

参考文献

[1] 张雪松, 李鹏, 周亦尧, 等. 基于贝叶斯概率的光伏出力组合预测方法[J]. 太阳能学报, 2021, 42(10): 80-86.
ZHANG X S, LI P, ZHOU Y Y, et al.Photovoltaic output combination forecasting method based on Bayesian probability[J]. Acta energiae solaris sinica, 2021, 42(10): 80-86.
[2] 杜晓东, 赵建利, 刘科研, 等. 基于数字孪生的光伏高比例配电网过载风险预警方法[J]. 电力系统保护与控制, 2022, 50(9): 136-144.
DU X D, ZHAO J L, LIU K Y, et al.Digital twin early warning method study for overload risk of distribution network with a high proportion of photovoltaic access[J]. Power system protection and control, 2022, 50(9): 136-144.
[3] 张立栋, 李继影, 吴颖, 等. 不同时间分辨率的风功率时间序列ARIMA模型预测[J]. 中国电力, 2016, 49(6): 176-180.
ZHANG L D, LI J Y, WU Y, et al.ARIMA model forecast for wind power time series with different temporal resolutions[J]. Electric power, 2016, 49(6): 176-180.
[4] 王献志, 曾四鸣, 周雪青, 等. 基于XGBoost联合模型的光伏发电功率预测[J]. 太阳能学报, 2022, 43(4): 236-242.
WANG X Z, ZENG S M, ZHOU X Q, et al.Power forecast of photovoltaic generation based on XGBoost combined model[J]. Acta energiae solaris sinica, 2022, 43(4): 236-242.
[5] 谭海旺, 杨启亮, 邢建春, 等. 基于XGBoost-LSTM组合模型的光伏发电功率预测[J]. 太阳能学报, 2022, 43(8): 75-81.
TAN H W, YANG Q L, XING J C, et al.Photovoltaic power prediction based on combined XGBoost-LSTM model[J]. Acta energiae solaris sinica, 2022, 43(8): 75-81.
[6] 杨楠, 周峥, 陈道君, 等. 基于非参数核密度估计的风功率波动性概率密度建模方法[J]. 太阳能学报, 2019, 48(7): 2028-2035.
YANG N, ZHOU Z, CHEN D J, et al.Research of modeling method based on non-parametric kernel density estimation of probability of wind power fluctuations[J]. Acta energiae solaris sinica, 2019, 48(7): 2028-2035.
[7] ZHANG W J, QUAN H, GANDHI O, et al.Improving probabilistic load forecasting using quantile regression NN with skip connections[J]. IEEE transactions on smart grid, 2020, 11(6): 5442-5450.
[8] CHAI S J, XU Z, WONG W K.Optimal granule-based PIs construction for solar irradiance forecast[J]. IEEE transactions on power systems, 2016, 31(4): 3332-3333.
[9] XIE J R, HONG T, LAING T, et al.On normality assumption in residual simulation for probabilistic load forecasting[J]. IEEE transactions on smart grid, 2017, 8(3): 1046-1053.
[10] LIN F, ZHANG Y, WANG K, et al.Parametric probabilistic forecasting of solar power with fat-tailed distributions and deep neural networks[J]. IEEE transactions on sustainable energy, 2022, 13(4): 2133-2147.
[11] 贾德香, 吕干云, 林芬, 等. 基于SAPSO-BP和分位数回归的光伏功率区间预测[J]. 电力系统保护与控制, 2021, 49(10): 20-26.
JIA D X, LYU G Y, LIN F, et al.Photovoltaic power interval prediction based on SAPSO-BP and quantile regression[J]. Power system protection and control, 2021, 49(10): 20-26.
[12] RUAS M, STREET A, FERNANDES C.A multi-quantile regression time series model with interquantile lipschitz regularization for wind power probabilistic forecasting[J]. Electric power systems research, 2022, 209: 107973.
[13] 程泽, 刘冲, 刘力. 基于相似时刻的光伏出力概率分布估计方法[J]. 电网技术, 2017, 41(2): 448-454.
CHENG Z, LIU C, LIU L.A method of probabilistic distribution estimation of PV generation based on similar time of day[J]. Power system technology, 2017, 41(2): 448-454.
[14] WANG K, ZHANG Y, LIN F, et al.Nonparametric probabilistic forecasting for wind power generation using quadratic spline quantile function and autoregressive recurrent neural network[J]. IEEE transactions on sustainable energy, 2022, 13(4): 1930-1943.
[15] WANG X Q, HYNDMAN R J, LI F, et al.Forecast combinations: an over 50-year review[J]. International journal of forecasting, 2023, 39(4): 1518-1547.
[16] 游文霞, 申坤, 杨楠, 等. 基于Bagging异质集成学习的窃电检测[J]. 电力系统自动化, 2021, 45(2): 105-113.
YOU W X, SHEN K, YANG N, et al.Electricity theft detection based on Bagging heterogeneous ensemble learning[J]. Automation of electric power systems, 2021, 45(2): 105-113.
[17] LI T Y, WANG Y, ZHANG N.Combining probability density forecasts for power electrical loads[J]. IEEE transactions on smart grid, 2020, 11(2): 1679-1690.
[18] WANG Y, ZHANG N, TAN Y S, et al.Combining probabilistic load forecasts[J]. IEEE transactions on smart grid, 2019, 10(4): 3664-3674.
[19] LU C B, LIANG J H, JIANG W Q, et al.High-resolution probabilistic load forecasting: a learning ensemble approach[J]. Journal of the Franklin Institute, 2023, 360(6): 4272-4296.
[20] LICHTENDAHL K C, WINKLER R L.Why do some combinations perform better than others?[J]. International journal of forecasting, 2020, 36(1): 142-149.