基于BP神经网络结合ERA5数据的风电功率预测

doi:10.19799/j.cnki.2095-4239.2024.0603

摘要/Abstract

摘要：

随着我国风力发电技术的不断发展和完善，风电在电力系统运行和调度的作用越来越突出。为了高效准确地预测风电功率，减少大量风电入网带来的负面影响，本文基于BP神经网络结合ERA5数据对我国北方某风电场进行风电功率预测，并采用粒子群优化(particle swarm algorithm，PSO)算法优化模型，结合平均绝对误差、均方根误差和Pearson相关系数分析风电功率预测效果。结果表明，模型训练集中预测与实测风电功率变化趋势基本一致，呈现同增同减的趋势，BP模型的平均绝对误差为702.12 W，均方根误差为1000.18 W，相关系数为0.91，PSO-BP模型的平均绝对误差为700.75 W，均方根误差为995.16 W，相关系数为0.94；测试集中ERA5数据在一定程度上高估了风电功率，但整体趋势基本一致，BP模型的平均绝对误差为861.09 W，均方根误差为1150.86 W，相关系数为0.81；PSO-BP模型的平均绝对误差为829.55 W，均方根误差为1117.39 W，相关系数为0.83，模型的预测效果相对较好，PSO-BP模型相较于BP模型的预测效果均有一定程度的提高，在该区域的风电功率预测方面有较好的适用性。研究结果可为缺乏观测数据或观测数据质量不高的地区预测风电功率提供参考。

关键词: 风力发电, BP神经网络, ERA5再分析资料, 粒子群优化算法, 风电功率预测

Abstract:

With the continuous development and improvement of wind power generation technology in China, its significance in power system operation and scheduling is becoming increasingly prominent. To predict wind power efficiently and accurately and reduce the negative impact of a large number of wind power on the grid, this paper focuses on a backpropagation (BP) neural network combined with ERA5 data for wind power prediction of a wind farm in the northern part of China and uses particle swarm optimization (PSO) algorithm to optimize the model, combined with the average absolute error, root mean square error and Pearson correlation coefficient to analyze the effect of wind power prediction. The results showed that the predicted and measured wind power changes in the model training set are the same, exhibiting a similar trend of increase and decrease. The average absolute error of the BP model is 702.12 W, with a root mean square error of 1000.18 W and the correlation coefficient is 0.91, whereas the average absolute error of the PSO-BP model is 700.75 W, with a root mean square error of 995.16 W, and the correlation coefficient is 0.94. The ERA5 data in the test set overestimates the wind power to a certain extent; however, the overall trend remains the same. The average absolute error of the BP model is 861.09W, with a root mean square error of 1150.86W, and the correlation coefficient is 0.81, whereas the average absolute error of the PSO-BP model of 829.55W, with a root mean square error of 1117.39 W, and the correlation coefficient is 0.83, indicating effective modeling. The prediction effect of the PSO-BP model is relatively good, and the PSO-BP model has a certain degree of improvement compared with the BP model, which has good applicability in predicting wind power in this region. The results of this study can provide a reference for the prediction of wind power in areas with limited observational data or where the quality of observational data is not high.

Key words: wind power, BP neural network, ERA5 reanalysis information, particle swarm algorithm, wind power prediction

中图分类号:

P4 68

王婷婷, 李斯胜, 于伟, 能锋田, 李星南, 杨佳琳, 熊亮. 基于BP神经网络结合ERA5数据的风电功率预测[J]. 储能科学与技术, 2025, 14(1): 183-189.

Tingting WANG, Sisheng LI, Wei YU, Fengtian NENG, Xingnan LI, Jialin YANG, Liang XIONG. Wind power prediction based on BP neural network combined with ERA5 data[J]. Energy Storage Science and Technology, 2025, 14(1): 183-189.

图/表 8

图1

图2

图3

表1

图4

表2

图5

表3

参考文献 14

1	ZHAO H R, WU Q W, HU S J, et al. Review of energy storage system for wind power integration support[J]. Applied Energy, 2015, 137: 545-553. DOI: 10.1016/j.apenergy.2014.04.103.
2	何玉灵, 焦凌钰, 孙凯, 等. 基于变权重组合的短期风光发电功率混合预测[J/OL]. 华北电力大学学报(自然科学版), 1-12 [2024-05-31]. http://kns.cnki.net/kcms/detail/13.1212.TM.20230914.1657.002.html.
	HE Y L, JIAO L Y, SUN K, et al. Short-term wind and photovoltaic power hybrid prediction based on variable weight combination prediction model[J/OL]. Journal of North China Electric Power University, 1-12 [2024-05-31]. http://kns.cnki.net/kcms/detail/13.1212.TM.20230914.1657.002.html.
3	WANG C H, ZHAO Q G, TIAN R. Short-term wind power prediction based on a hybrid Markov-based PSO-BP neural network[J]. Energies, 2023, 16(11): 4282. DOI: 10.3390/en16114282.
4	ZHANG Y H, LI P, LI H X, et al. Short-term power prediction of wind power generation system based on logistic chaos atom search optimization BP neural network[J]. International Transactions on Electrical Energy Systems, 2023, 2023: 6328119. DOI: 10.1155/2023/6328119.
5	贺莉微, 孙朋杰, 曾琦. ERA5再分析资料在风能资源方面的应用[J]. 湖北农业科学, 2021, 60(24): 70-75. DOI: 10.14088/j.cnki.issn0439-8114.2021.24.016.
	HE L W, SUN P J, ZENG Q. Application of ERA5 reanalysis data in wind energy resources[J]. Hubei Agricultural Sciences, 2021, 60(24): 70-75. DOI: 10.14088/j.cnki.issn0439-8114.2021.24.016.
6	周正斌, 张艺丹, 罗坤, 等. 基于高分辨率再分析资料的西南地区风资源特征分析[J]. 成都信息工程大学学报, 2023, 38(1): 75-82. DOI: 10.16836/j.cnki.jcuit.2023.01.012.
	ZHOU Z B, ZHANG Y D, LUO K, et al. Analysis of wind resources in Southwest China based on high-resolution reanalysis data[J]. Journal of Chengdu University of Information Technology, 2023, 38(1): 75-82. DOI: 10.16836/j.cnki.jcuit.2023.01.012.
7	ZHAI R W, HUANG C J, YANG W, et al. Applicability evaluation of ERA5 wind and wave reanalysis data in the South China Sea[J]. Journal of Oceanology and Limnology, 2023, 41(2): 495-517. DOI: 10.1007/s00343-022-2047-8.
8	ZEKEIK Y, ORTIZBEVIA M J, ALVAREZ-GARCIA F J, et al. Long-term assessment of Morocco's offshore wind energy potential using ERA5 and IFREMER wind data[J]. Journal of Marine Science and Engineering, 2024, 12(3): 460. DOI: 10.3390/jmse12030460.
9	陈君来, 张周文, 陈蔚华, 等. ERA5再分析数据在中小型水库水位模拟的适用性分析——以长江水库为例[J]. 人民珠江, 2023, 44(12): 127-135.
	CHEN J L, ZHANG Z W, CHEN W H, et al. Applicability of ERA5 reanalysis data in simulating water level of medium and small-scale reservoirs: A case study of Changjiang reservoir[J]. Pearl River, 2023, 44(12): 127-135.
10	BELMONTE RIVAS M, STOFFELEN A. Characterizing ERA-Interim and ERA5 surface wind biases using ASCAT[J]. Ocean Science, 2019, 15(3): 831-852. DOI: 10.5194/os-15-831-2019.
11	黄凤新, 刘寿东, 祝赢, 等. 基于滚动极值处理的BP神经网络方法的WRF模式预报风速订正[J]. 科学技术与工程, 2013, 13(7): 1768-1772. DOI: 10.3969/j.issn.1671-1815.2013.07.012.
	HUANG F X, LIU S D, ZHU Y, et al. Application of BP neural network based on rolling extremal management in revising wind speed forecasting of WRF model[J]. Science Technology and Engineering, 2013, 13(7): 1768-1772. DOI: 10.3969/j.issn.1671-1815.2013.07.012.
12	李晓娟, 张芳媛, 喻玲. 基于主成分分析-BP神经网络的风电备件需求预测[J]. 科学技术与工程, 2024, 24(1): 281-288.
	LI X J, ZHANG F Y, YU L. Wind power spare parts demand forecasting based on PCA-BP neural network[J]. Science Technology and Engineering, 2024, 24(1): 281-288.
13	史博明. 基于组合模型的短期风电功率预测研究[D]. 石家庄: 河北科技大学, 2021. DOI: 10.27107/d.cnki.ghbku.2021.000448.
	SHI B M. Research on short-term wind power forecasting based on combined model[D]. Shijiazhuang: Hebei University of Science and Technology, 2021. DOI: 10.27107/d.cnki.ghbku. 2021.000448.
14	逯登龙, 高鹏, 范丽锋, 等. 基于GA- BP神经网络的风电功率预测方法研究[J]. 自动化仪表, 2024, 45(3): 97-102. DOI: 10.16086/j.cnki.issn1000-0380.2022030096.
	LU D L, GAO P, FAN L F, et al. Research on wind power prediction method based on GA-BP neural network[J]. Process Automation Instrumentation, 2024, 45(3): 97-102. DOI: 10.16086/j.cnki.issn1000-0380.2022030096.

神经元数量	均方根误差
3	1027.3
4	1016.0
5	1009.3
6	1001.3
7	998.3
8	1011.6
9	1000.3
10	990.3
11	984.9
12	990.9

[1]	赵鑫浩, 许亮. 改进的萤火虫算法优化反向传播神经网络动力锂离子电池健康状态估计[J]. 储能科学与技术, 2023, 12(3): 934-940.
[2]	杨帆, 和嘉睿, 陆鸣, 陆玲霞, 于淼. 基于BP-UKF算法的锂离子电池SOC估计[J]. 储能科学与技术, 2023, 12(2): 552-559.
[3]	刘芊彤, 邢远秀. 基于VMD-PSO-GRU模型的锂离子电池剩余寿命预测[J]. 储能科学与技术, 2023, 12(1): 236-246.
[4]	耿萌萌, 范茂松, 杨凯, 赵光金, 谭震, 高飞, 张明杰. 基于EIS和神经网络的退役电池SOH快速估计[J]. 储能科学与技术, 2022, 11(2): 673-678.
[5]	曾伟, 熊俊杰, 马速良, 谭宇良, 李建林. 提升光伏消纳的分布式储能系统控制方法[J]. 储能科学与技术, 2022, 11(10): 3268-3274.
[6]	王帅, 马鸿雁, 窦嘉铭, 张英达, 李晟延, 胡璐锦. 基于UGOA-BP的锂电池SOC估算[J]. 储能科学与技术, 2022, 11(1): 258-264.
[7]	陈星宇, 刘忠, 寇攀高, 邹淑云, 潘宜桦. 离网型风-光-抽水蓄能恒压供电系统[J]. 储能科学与技术, 2021, 10(1): 355-361.
[8]	张远进, 吴华伟, 叶从进. 基于AUKF-BP神经网络的锂电池SOC估算[J]. 储能科学与技术, 2021, 10(1): 237-241.
[9]	王泰华, 张书杰, 陈金干. 基于BP-PSO算法的锂电池低温充电策略优化[J]. 储能科学与技术, 2020, 9(6): 1940-1947.
[10]	范永生, 赵璐璐, 刘庆华, 缪平. 陆上风场液流电池储能经济性分析[J]. 储能科学与技术, 2020, 9(3): 725-729.
[11]	张新锋, 姚蒙蒙, 王钟毅, 饶勇翔. 基于ACO-BP神经网络的锂离子电池容量衰退预测[J]. 储能科学与技术, 2020, 9(1): 138-144.
[12]	吴华伟, 张远进, 叶从进. 基于萤火虫神经网络的动力电池SOC估算[J]. 储能科学与技术, 2019, 8(3): 575-579.
[13]	戴兴建, 魏鲲鹏, 张小章, 姜新建, 张剀. 飞轮储能技术研究五十年评述[J]. 储能科学与技术, 2018, 7(5): 765-782.
[14]	韩永杰, 任正义, 吴滨, 李翀. 飞轮储能系统在1.5 MW风机上的应用研究[J]. 储能科学与技术, 2015, 4(2): 198-202.
[15]	刘颖明, 徐中民, 王晓东. 考虑飞轮储能的风电场有功功率平滑控制[J]. 储能科学与技术, 2015, 4(2): 194-197.