基于DALSTM和联合分位数损失的海上风电功率概率预测

doi:10.11930/j.issn.1004-9649.202212011

中国电力 ›› 2023, Vol. 56 ›› Issue (11): 10-19.DOI: 10.11930/j.issn.1004-9649.202212011

• 海上风电送出与并网技术 • 上一篇下一篇

基于DALSTM和联合分位数损失的海上风电功率概率预测

苏向敬¹(), 宇海波¹(), 符杨¹(), 田书欣¹, 李海瑜², 耿福海²

1. 上海电力大学电气工程学院，上海　200090
2. 国家电投集团上海能源科技发展有限公司，上海　200233

收稿日期:2022-12-05 接受日期:2023-05-04 出版日期:2023-11-28 发布日期:2023-11-28
作者简介:苏向敬（1984—），男，博士，副教授，从事海上风电大数据技术、主动配电网优化规划运行研究，E-mail： xiangjing_su@126.com
宇海波（1996—），男，硕士研究生，从事基于人工智能的海上风电功率概率预测技术研究，E-mail： 1014565348@qq.com
符杨（1968—），男，通信作者，博士，教授，博士生导师，从事风力发电与并网技术、变压器状态检测与故障诊断研究，E-mail： mfudong@126.com
基金资助:
国家自然科学基金资助项目（基于Vague软集的海上风电集群组网接入下输电网鲁棒扩展规划方法研究，52007112）；上海市教育委员会科研创新计划资助项目（面向大规模海上风电友好接入的海上电网规划与优化运行理论方法，2021-01-07-00-07-E00122）。

Probabilistic Forecasting of Offshore Wind Power Based on Dual-stage Attentional LSTM and Joint Quantile Loss Function

Xiangjing SU¹(), Haibo YU¹(), Yang FU¹(), Shuxin TIAN¹, Haiyu LI², Fuhai GENG²

1. School of Electrical Engineering, Shanghai University of Electric Power, Shanghai 200090, China
2. Shanghai Energy Technology Development Co. Ltd., State Power Investment Corporation Limited, Shanghai 200233, China

Received:2022-12-05 Accepted:2023-05-04 Online:2023-11-28 Published:2023-11-28
Supported by:
This work is supported by National Natural Science Foundation of China (Research on Robust Expansion Planning Method of Offshore Wind Power Cluster Network Access to the Lower Transmission Grid Based on Vague Soft Set, No.52007112) and Scientific Research & Innovation Program of Shanghai Education Commission (Theoretical Methods for the Planning and Optimized Operation of Offshore Power Grids Friendly to Large-Scale Offshore Wind Power Integration, No.2021-01-07-00-07-E00122).

摘要/Abstract

摘要：

传统特征关联方法的预设阈值限制及分位数损失中各分位点损失的量级差异，使得海上风电功率概率预测精度受限。为了提高概率预测精度，提出了一种基于多任务联合分位数损失的双重注意力概率预测模型（MT-DALSTM）。首先，引入特征和时序双重注意力机制对特征间的关联关系和时序依赖性进行挖掘，赋予关键特征和时间点信息以注意力权重来提升功率预测的准确性；其次，在模型训练方面，采用一种基于任务不确定性的多任务联合分位数损失，通过动态调整各损失权重占比来提升最终预测结果的综合性能指标；最后，基于东海大桥海上风电场真实数据仿真验证结果表明：相比于现有的风电概率预测研究，所提方法在锐度、可靠性、综合性能指标上均具有明显提升，验证了该模型提高预测精度的有效性。

关键词: 海上风电, 概率预测, 注意力机制, 注意力权重, 特征关联性, 分位数回归

Abstract:

Probabilistic prediction of offshore wind power is not high in accuracy due to the predetermined threshold limitation of the traditional feature correlation method and the magnitude difference of the quantile loss in each quantile loss. To improve the probabilistic prediction accuracy, a multi-task joint quantile loss-based dual-attention probabilistic prediction model (MT-DALSTM) is proposed. Firstly, a feature and temporal dual attention mechanism is introduced to mine the correlation and temporal dependence among features, and attention weights are given to key features and time point information to improve the accuracy of power prediction. Secondly, during model training, the multi-task joint quantile loss based on task uncertainty is used to improve the final prediction results by dynamically adjusting the proportion of each loss weight. Finally, the simulation validation results based on the real data from the Donghai Bridge offshore wind farm show that the proposed method has significant improvement in sharpness, reliability and comprehensive performance indexes compared to the existing wind power probabilistic prediction studies, which verifies the effectiveness of the model in improving the prediction accuracy.

Key words: offshore wind power, probabilistic prediction, attention mechanism, attention weight, feature relevance, quantile regression

苏向敬, 宇海波, 符杨, 田书欣, 李海瑜, 耿福海. 基于DALSTM和联合分位数损失的海上风电功率概率预测[J]. 中国电力, 2023, 56(11): 10-19.

Xiangjing SU, Haibo YU, Yang FU, Shuxin TIAN, Haiyu LI, Fuhai GENG. Probabilistic Forecasting of Offshore Wind Power Based on Dual-stage Attentional LSTM and Joint Quantile Loss Function[J]. Electric Power, 2023, 56(11): 10-19.

导出引用管理器 EndNote|Ris|BibTeX

链接本文: https://www.electricpower.com.cn/CN/10.11930/j.issn.1004-9649.202212011

https://www.electricpower.com.cn/CN/Y2023/V56/I11/10

图/表 13

图 1 LSTM结构

Fig.1 LSTM structure

图 2 特征注意力结构

Fig.2 Feature attention structure

图 3 时序注意力结构

Fig.3 Temporal attention structure

图 4 分位数预测结构

Fig.4 Quantile forecasting structure

图 5 预测流程

Fig.5 Forecast flow chart

图 6 不同模型有功单步预测结果

Fig.6 Power predictions by different models

表 1 概率预测结果对比

Table 1 Comparison of probabilistic predicting results

预测模型	S_AW/kW	S_CRPS
DQR	483.23	94.94
LSTM-QR	457.44	89.54
DALSTM	407.28	71.41
MT-DALSTM	372.37	64.23

图 7 不同模型预测的平均覆盖率误差

Fig.7 ACD of different models

图 8 不同模型多步预测的连续等级概率分数

Fig.8 CRPS of different models

图 9 特征注意力权重

Fig.9 Feature attention weights

图 10 变桨特征对比

Fig.10 Comparison of pitch characteristics

表 2 变桨特征预测结果对比

Table 2 Comparison of pitch characteristics probabilistic predicting results

预测模型	S_AW/kW	S_CRPS
MT-DALSTM（无变桨特征）	394.46	69.17
MT-DALSTM	372.37	64.23

图 11 时序注意力权重

Fig.11 Temporal attention weights

参考文献 25

1	全球风能协会(GWEC). 全球风能2022年度报告[EB/OL].https://gwec. net/policy- research/reports/.
2	孙荣富, 张涛, 和青, 等. 风电功率预测关键技术及应用综述[J]. 高电压技术, 2021, 47 (4): 1129- 1143. DOI
	SUN Rongfu, ZHANG Tao, HE Qing, et al. Review on key technologies and applications in wind power forecasting[J]. High Voltage Engineering, 2021, 47 (4): 1129- 1143. DOI
3	迟永宁, 梁伟, 张占奎, 等. 大规模海上风电输电与并网关键技术研究综述[J]. 中国电机工程学报, 2016, 36 (14): 3758- 3771. DOI
	CHI Yongning, LIANG Wei, ZHANG Zhankui, et al. An overview on key technologies regarding power transmission and grid integration of large scale offshore wind power[J]. Proceedings of the CSEE, 2016, 36 (14): 3758- 3771. DOI
4	YUAN X H, TAN Q X, LEI X H, et al. Wind power prediction using hybrid autoregressive fractionally integrated moving average and least square support vector machine[J]. Energy, 2017, 129, 122- 137. DOI
5	LIU X L, LIN Z, FENG Z M. Short-term offshore wind speed forecast by seasonal ARIMA - a comparison against GRU and LSTM[J]. Energy, 2021, 227, 120492. DOI
6	张群, 唐振浩, 王恭, 等. 基于长短时记忆网络的超短期风功率预测模型[J]. 太阳能学报, 2021, 42 (10): 275- 281. DOI
	ZHANG Qun, TANG Zhenhao, WANG Gong, et al. Ultra-short-term wind power prediction model based on long and short term memory network[J]. Acta Energiae Solaris Sinica, 2021, 42 (10): 275- 281. DOI
7	LIN Z, LIU X L. Wind power forecasting of an offshore wind turbine based on high-frequency SCADA data and deep learning neural network[J]. Energy, 2020, 201, 117693. DOI
8	黄冬梅, 庄兴科, 胡安铎, 等. 基于灰色关联分析和K均值聚类的短期负荷预测[J]. 电力建设, 2021, 42 (7): 110- 117. DOI
	HUANG Dongmei, ZHUANG Xingke, HU Anduo, et al. Short-term load forecasting based on similar-day selection with GRA-K-means[J]. Electric Power Construction, 2021, 42 (7): 110- 117. DOI
9	梁智, 孙国强, 卫志农, 等. 基于变量选择与高斯过程回归的短期负荷预测[J]. 电力建设, 2017, 38 (2): 122- 128. DOI
	LIANG Zhi, SUN Guoqiang, WEI Zhinong, et al. Short-term load forecasting based on variable selection and Gaussian process regression[J]. Electric Power Construction, 2017, 38 (2): 122- 128. DOI
10	MNIH V, HEESS N, GRAVES A, et al. Recurrent models of visual attention[EB/OL]. 2014: arXiv: 1406.6247.https://arxiv.org/abs/1406.6247.
11	YUAN K, ZHANG K F, ZHENG Y X, et al. Irregular distribution of wind power prediction[J]. Journal of Modern Power Systems and Clean Energy, 2018, 6 (6): 1172- 1180. DOI
12	WANG Y, HU Q H, MENG D Y, et al. Deterministic and probabilistic wind power forecasting using a variational Bayesian-based adaptive robust multi-kernel regression model[J]. Applied Energy, 2017, 208, 1097- 1112. DOI
13	WANG Z, WANG W S, LIU C, et al. Short-term probabilistic forecasting for regional wind power using distance-weighted kernel density estimation[J]. IET Renewable Power Generation, 2018, 12 (15): 1725- 1732. DOI
14	阎洁, 刘永前, 韩爽, 等. 分位数回归在风电功率预测不确定性分析中的应用[J]. 太阳能学报, 2013, 34 (12): 2101- 2107. DOI
	YAN Jie, LIU Yongqian, HAN Shuang, et al. Quantile regression in uncertainty analysis of wind power forecasting[J]. Acta Energiae Solaris Sinica, 2013, 34 (12): 2101- 2107. DOI
15	CANNON A J. Quantile regression neural networks: implementation in R and application to precipitation downscaling[J]. Computers & Geosciences, 2011, 37 (9): 1277- 1284.
16	WAN C, LIN J, WANG J H, et al. Direct quantile regression for nonparametric probabilistic forecasting of wind power generation[J]. IEEE Transactions on Power Systems, 2017, 32 (4): 2767- 2778. DOI
17	WANG H Z, WANG G B, LI G Q, et al. Deep belief network based deterministic and probabilistic wind speed forecasting approach[J]. Applied Energy, 2016, 182, 80- 93. DOI
18	李丹, 张远航, 杨保华, 等. 基于约束并行LSTM分位数回归的短期电力负荷概率预测方法[J]. 电网技术, 2021, 45 (4): 1356- 1364. DOI
	LI Dan, ZHANG Yuanhang, YANG Baohua, et al. Short time power load probabilistic forecasting based on constrained parallel-LSTM neural network quantile regression mode[J]. Power System Technology, 2021, 45 (4): 1356- 1364. DOI
19	GUO T, LIN T, ANTULOV-FANTULIN N. Exploring interpretable LSTM neural networks over multi-variable data[EB/OL]. 2019: arXiv: 1905.12034.https://arxiv.org/abs/1905.12034.
20	LI A, XIAO F, ZHANG C, et al. Attention-based interpretable neural network for building cooling load prediction[J]. Applied Energy, 2021, 299, 117238. DOI
21	HE X N, HE Z K, SONG J K, et al. NAIS: neural attentive item similarity model for recommendation[J]. IEEE Transactions on Knowledge and Data Engineering, 2018, 30 (12): 2354- 2366. DOI
22	HU J M, TANG J W, LIN Y Y. A novel wind power probabilistic forecasting approach based on joint quantile regression and multi-objective optimization[J]. Renewable Energy, 2020, 149, 141- 164. DOI
23	CIPOLLA R, GAL Y, KENDALL A. Multi-task learning using uncertainty to weigh losses for scene geometry and semantics[C]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Salt Lake City, UT, USA. IEEE, 2018: 7482–7491.
24	万灿, 崔文康, 宋永华. 新能源电力系统概率预测: 基本概念与数学原理[J]. 中国电机工程学报, 2021, 41 (19): 6493- 6509.
	WAN Can, CUI Wenkang, SONG Yonghua. Probabilistic forecasting for power systems with renewable energy sources: basic concepts and mathematical principles[J]. Proceedings of the CSEE, 2021, 41 (19): 6493- 6509.
25	WAN C, NIU M, SONG Y H, et al. Pareto optimal prediction intervals of electricity price[J]. IEEE Transactions on Power Systems, 2017, 32 (1): 817- 819. DOI

[1]	丰力, 张莲梅, 韦家佳, 邓长虹, 李果, 尹家悦. 基于全生命周期经济评估的海上风电发展与思考[J]. 中国电力, 2024, 57(9): 80-93.
[2]	姜文瑾, 刘巧妹, 杨晓东, 阙定飞, 沈豫, 黄夏楠, 赖振华. 计及气固两相储氢特性的海上风电-多元储能系统优化配置[J]. 中国电力, 2024, 57(9): 103-112.
[3]	黄宁泊, 高建伟, 许传博, 徐选华, 赵舒通, 缑迅杰, 姜晓静. 基于经验挖掘与混合语言的海上风电制氢加氢港口选址研究[J]. 中国电力, 2024, 57(9): 113-123.
[4]	叶婧, 蔡俊文, 张磊, 周广浩, 何杰辉, 翟学. 考虑海缆实际载流量的海上风电集电系统拓扑优化[J]. 中国电力, 2024, 57(7): 173-181.
[5]	陈中飞, 赵越, 蔡秋娜, 张乔榆, 王泽林, 戴晓娟, 陈雨果. 基于净负荷预测误差统计的电力系统爬坡能力充裕度评估[J]. 中国电力, 2024, 57(5): 50-60.
[6]	高岩, 吴汉斌, 张纪欣, 张华铭, 张沛. 基于组合深度学习的光伏功率日前概率预测模型[J]. 中国电力, 2024, 57(4): 100-110.
[7]	阎洁, 杨佳琳, 王航宇, 卢姣阳, 刘永前, 张磊. 基于风况预测误差自适应的海上风电场尾流偏转控制方法[J]. 中国电力, 2024, 57(3): 190-196.
[8]	陈璨, 苏紫诺, 马原, 刘佳林, 王玉庆, 王飞. 基于分层关联性建模的分布式光伏功率超短期概率预测[J]. 中国电力, 2024, 57(12): 50-59.
[9]	吴晓刚, 阎洁, 葛畅, 唐雅洁, 倪筹帷, 季青锋. 基于改进GRU-CNN的风光水一体化超短期功率预测方法[J]. 中国电力, 2023, 56(9): 178-186,205.
[10]	叶婧, 周广浩, 张磊, 杨莉, 翟学, 蔡俊文. 考虑馈线交叉规避的海上风电场海缆路径优化[J]. 中国电力, 2023, 56(6): 167-175.
[11]	苏开元, 董文凯, 邱银锋, 魏澈, 谢小荣. 分散式风储一体化系统提升海上油田群电网频率稳定性研究[J]. 中国电力, 2023, 56(5): 163-171.
[12]	杨舒婷, 陈新, 黄通, 魏麒璇. 考虑MMC环流控制的海上风电经柔直送出系统阻抗塑造方法[J]. 中国电力, 2023, 56(4): 38-45.
[13]	陆友文, 崔昊, 陈佳宁, 彭祥佳, 冯双, 刘栋. 基于RA-CNN和同步相量的风电场次/超同步振荡参数智能辨识方法[J]. 中国电力, 2023, 56(4): 46-55,67.
[14]	王朝辉, 黄松阁, 林斌, 陈雨薇, 范淑敏, 施朝晖. 基于海上风电场区66 kV集电系统海缆最大截面经济性分析[J]. 中国电力, 2023, 56(11): 20-28.
[15]	周震震, 宋云海, 何宇浩, 王黎伟, 黄和燕, 何珏, 朱志航, 闫云凤. 基于分组查询注意力的可扩展电力人员行为分类方法[J]. 中国电力, 2023, 56(11): 77-85.

基于DALSTM和联合分位数损失的海上风电功率概率预测

Probabilistic Forecasting of Offshore Wind Power Based on Dual-stage Attentional LSTM and Joint Quantile Loss Function

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 13

参考文献 25

相关文章 15

编辑推荐

Metrics

模态框（Modal）标题

基于DALSTM和联合分位数损失的海上风电功率概率预测

Probabilistic Forecasting of Offshore Wind Power Based on Dual-stage Attentional LSTM and Joint Quantile Loss Function

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 13

参考文献 25

相关文章 15

编辑推荐

Metrics