Application of Ensemble Selection Method in Short-term Wind Power Forecasting
-
摘要: 为提高短期风速及功率预测的准确率，减小风电不确定性对电网系统的影响，尝试利用预测窗口期的风速观测进行数值天气预报的集合成员选优，挑选和实际风速更接近的相似预报成员，并构成选优集合进行机器学习模型的训练和测试。相较仅使用集合平均的常规方法，该方法考虑了不同集合成员之间的预报差异，避免了引入误差较大的集合成员，从而有利于改善预报风速偏差。利用不同海拔高度、不同地形特征的河南、甘肃两个风电场中不同集合的表现及敏感性试验结果，确定风电场最佳选优集合数量。相较于集合平均的结果，集合选优方案在不同天气过程中能较好地预报风速的起降，与实际风速更接近，且海平面气压场整体更接近ERA5。对不同风电场进行连续十一个月的风速及功率预测对比试验，结果表明，集合选优方法预报的风速日变化形态和月均风速较原集合平均方法均有改善。分析两个风场不同时长范围、不同速率变化的上坡风和下坡风观测数据可知，在0~2 h及2~4 h内，风速变化为2~4 m/s的个例最多。对比集合平均结果，集合选优方案对于该类型上、下坡风的预测精度均有较为明显的提升。利用机器学习算法对选优集合预报进行训练，能进一步降低风速的绝对偏差和均方根误差，从而有效改善功率预测精度。Abstract: To improve the accuracy of short-term wind speed and power forecasts as well as reduce the impact of wind power uncertainty on the grid system, this study attempted to use wind speed observations to select the optimal numerical forecasting ensemble members that closely matched the actual wind speed within the forecast window. Those selected ensemble members then formed an optimized ensemble for training and testing machine learning models. Unlike the conventional method that relied solely on ensemble averaging, this method considered the forecast discrepancies among different ensemble members, avoided the introduction of members with large errors, and thus helped to improve wind speed prediction. The optimal number of ensemble members for wind farms with different altitudes and terrains in Henan and Gansu was determined based on the results of ensemble performance and sensitivity experiments. Comparative analyses demonstrated that the ensemble selection forecasts outperformed ensemble averaging in predicting wind speed fluctuations during different weather processes, closely aligning with actual wind speed observations. The sea-level pressure field estimates generated by the ensemble selection method exhibited a higher level of agreement with ERA5 data. Evaluation of wind speed and power prediction over eleven consecutive months in different wind farms showed that the ensemble selection method improved the accuracy of diurnal variation and monthly average wind speed compared to the original ensemble averaging method. Analysis of the observed data of upslope and downslope winds with different durations and speed changes in two wind farms showed that there were the most wind speed changes of 2—4 m s-1 within 0—2 h and 2—4 h. Compared to ensemble averaging, the ensemble selection method significantly improved the prediction accuracy for these upslope and downslope winds. Furthermore, using the machine learning algorithm to train the optimal selection ensemble can further reduce the absolute deviation and root mean square error of wind speed, thereby effectively improving the accuracy of power prediction.
-
图 5 2021年6月19日20时、7月27日20时及10月2日20时三次预报过程河南风电场实际风速（黑线）、集合平均风速（蓝线）和集合选优风速（红线）的时序变化（阴影表示集合预报区间）（a、d、g），及ERA5与集合平均（b、e、h）、集合选优（c、f、i）海平面气压平均差异场（单位：Pa，黑框表示风场区域）
图 7 实际风速（黑实线）、集合平均风速（蓝实线）、集合选优风速（红实线）的月平均风速变化及月实际风速日变化与年风速日变化的相关系数（虚线）
a.河南风电场; b.甘肃风电场。
图 9 河南及甘肃风电场使用不同方案风速预报的逐月均方根误差和绝对偏差（单位：m · s-1）
a.河南风电场均方根误差; b.甘肃风电场均方根误差; c. 河南风电场绝对偏差; d. 甘肃风电场绝对偏差。
表 1 风电场逐月风速数据完整率
月份 河南风电场 甘肃风电场 4月 100.00% 81.20% 5月 99.70% 99.90% 6月 100.00% 100.00% 7月 82.50% 100.00% 8月 100.00% 100.00% 9月 100.00% 98.30% 10月 98.30% 93.80% 11月 100.00% 99.90% 12月 74.30% 87.90% 1月 67.90% 91.80% 2月 100.00% 77.40% 平均 93.00% 93.70% 
下载: 导出CSV
表 2 风速预测评估指标
指标 公式 说明 MAE $\text { MAE }=\frac{1}{N} \sum\limits_{i=1}^n\left|e_i\right|$ 平均绝对误差，其中n表示预报时刻点，N表示预报时刻总数，ei表示观测和预报的偏差 STD $\mathrm{STD}=\sqrt{\frac{1}{N} \sum\limits_{i=1}^n\left(e_i-\overline{e_N}\right)^2}$ 标准差，其中$\overline{e_N}$表示观测和预报在所有时刻的平均偏差，其余变量含义同上 RMSE $\text { RMSE }=\sqrt{\frac{1}{N} \sum\limits_{i=1}^n\left(e_i\right)^2}$ 均方根误差，式中各变量含义同上 COEF $\mathrm{COEF}=\frac{\sum\limits_{i=1}^n\left(x_i-\overline{x_N}\right)\left(y_i-\overline{y_N}\right)}{\sqrt{\sum\limits_{i=1}^n\left(x_i-\overline{x_N}\right)} \sqrt{\sum\limits_{i=1}^n\left(y_i-\overline{y_N}\right)}}$ 简单相关系数，其中xi和yi分别表示预报时刻i的预报值和观测值，$\overline{x_N}$和$\overline{y_N}$分别表示所有预报时刻点的预报均值与观测均值
下载: 导出CSV
表 3 河南及甘肃风电场不同集合选优数方案的评估结果
方案 RMSE MAE STD COEF ens_avg_henan 204.60% 161.30% 1.987 0.621 ens_sel5_henan 194.50% 152.60% 1.894 0.662 ens_sel15_henan 193.90% 152.50% 1.886 0.663 ens_sel25_henan 195.40% 153.60% 1.901 0.658 ens_sel30_henan 195.80% 153.90% 1.904 0.657 ens_avg_gansu 209.40% 161.70% 2.057 0.813 ens_sel5_gansu 206.90% 158.00% 2.042 0.821 ens_sel15_gansu 203.20% 156.10% 1.998 0.827 ens_sel25_gansu 201.70% 155.30% 1.979 0.831 ens_sel30_gansu 202.10% 155.70% 1.982 0.83
下载: 导出CSV
表 4 河南及甘肃风电场使用不同方案的风速及功率预测评估结果
方案 风速预测评估指标 功率预测评估指标 RMSE MAE COEF RMSE MAE COEF ACC ens_avg_henan 204.00% 159.00% 0.592 27.99 19.89 0.528 72% ens_sel_henan 194.00% 154.00% 0.635 26.19 19.29 0.575 74% ens_avg_svr_henan 190.00% 152.00% 0.562 23.64 16.05 0.487 76% ens_sel_svr_henan 183.00% 149.00% 0.603 22.01 15.6 0.531 78% ens_avg_gansu 207.00% 159.00% 0.818 80.54 53.77 0.796 80% ens_sel_gansu 199.00% 153.00% 0.832 78.05 51.87 0.813 81% ens_avg_svr_gansu 200.00% 149.00% 0.821 73.18 47.03 0.798 82% ens_sel_svr_gansu 189.00% 142.00% 0.836 70.08 44.95 0.817 83%
下载: 导出CSV
-
[1] 韩肖清, 李廷钧, 张东霞, 等. 双碳目标下的新型电力系统规划新问题及关键技术[J]. 高电压技术, 2021: 1-12. [2] Wind Energy Data[DB/OL]. https://www.irena.org/wind .[3] 康重庆, 姚良忠. 高比例可再生能源电力系统的关键科学问题与理论研究框架[J]. 电力系统自动化, 2017, 41(9): 2-11. [4] HAN S, ZHANG L N, LIU Y Q, et al. Quantitative evaluation method for the complementarity of wind-solar-hydro power and optimization of wind-solar ratio[J]. Applied energy, 2019, 236: 973-984. [5] 朱凌志, 陈宁, 韩华玲. 风电消纳关键问题及应对措施分析[J]. 电力系统自动化, 2011, 35(22): 29-34. [6] 薛禹胜, 郁琛, 赵俊华, 等. 关于短期及超短期风电功率预测的评述[J]. 电力系统自动化, 2015, 39(6): 141-151. [7] 丁明, 张超, 王勃等. 基于功率波动过程的风电功率短期预测及误差修正[J]. 电力系统自动化, 2019, 43(3): 2-12. [8] HAN S, ZHANG L N, LIU Y Q, et al. A data sample division method for wind power prediction based on China's 24 solar terms[J]. International Transactions on Electrical Energy Systems, 2020, 30(7): e12342. [9] 施涛, 李春来, 朱慧敏. 考虑功率预测不确定性的风电消纳随机调度[J]. 电网与清洁能源, 2019, 35(4): 55-59. [10] 黄树帮, 陈耀, 金宇清. 碳中和背景下多通道特征组合超短期风电功率预测[J]. 发电技术, 2021, 42(1): 60-68. [11] 胡佳琳, 赵翠妹, 肖成, 等. 计及发电权交易和碳交易的含风电电力系统低碳经济调度研究[J]. 电力与能源进展, 2018, 6(1): 54-66. [12] 彭小圣, 熊磊, 劲宇, 等. 风电集群短期及超短期功率预测精度改进方法综述[J]. 中国电机工程学报, 2016, 36(23): 6 315-6 326. [13] 杨剑南. 基于人工神经网络的短期风功率预测研究[D]. 河北: 华北电力大学, 2015. [14] CHANG G W, LU H J, ChANG Y R, et al. An improved neural network-based approach for short-term wind speed and power forecast[J]. Renewable Energy, 2017, 105(5): 301-311. [15] YANG D Z, WANG W T, HONG T. A historical weather forecast dataset from the European Centre for Medium-Range Weather Forecasts (ECMWF) for energy forecasting[J]. Solar Energy, 2022, 232: 263-274. [16] NIELSEN H A, MADSEN H, NIELSEN T S, et al. Wind Power Ensemble Forecasting Using Wind Speed and Direction Ensembles from ECMWF or NCEP[R /OL]. [2005]. https://www.researchgate.net/publication/233400558_Wind_Power_Ensemble_Forecasting_Using_Wind_Speed_and_Direction_Ensembles_from_ECMWF_or_NCEP [17] GIEBEL G, BADGER J, LANDBERG L, et al. Wind Power Prediction using Ensembles[J]. Campus Ris, 2005. [18] FOLEY A M, LEAHY P G, MARVUGLIA A, et al. Current methods and advances in forecasting of wind power generation[J]. Renewable Energy, 2012, 37(1): 1-8. [19] MONACHE L D, ECKEL F A, RIFE D L, et al. Probabilistic Weather Prediction with an Analog Ensemble[J]. Mon Wea Rev, 2013, 141 (10): 3 498-3 516. [20] 赵华生, 黄小燕, 黄颖. ECMWF集合预报产品在广西暴雨预报中的释用[J]. 应用气象学报, 2018, 29(3): 344-353. [21] 胡海川, 赵伟, 董林. 概率密度匹配方法在我国近海海面10 m风速预报中的应用[J]. 热带气象学报, 2021, 37(1): 91-101 [22] MUREAU R, MOLTENI F, PALMER T N. Ensemble prediction using dynamically conditioned perturbations[J]. Quart J Roy Meteor Soc, 2010, 119(510): 299-323. [23] TOTH Z, KALNAY E. Ensemble Forecasting at NMC: The Generation of Perturbations[J]. Bull Amer Meteor Soc, 1993, 74(12): 2 317-2 330 [24] CHEN Y D, GUO S, MENG D M, et al. The impact of optimal selected historical forecasting samples on hybrid ensemble-variational data assimilation[J]. Atmospheric Research, 2020, 242: 104980. [25] 陈耀登, 郭闪, 王元兵, 等. 基于优化样本组合的集合-变分混合同化方案研究[J]. 热带气象学报, 2020, 36(4): 464-476. [26] 赵文婧, 李照荣, 王小勇, 等. 相似误差订正方法在风电短期风速预报中的应用研究[J]. 热带气象学报, 2021, 37(1): 73-81. DOI: 10.16032/j.issn.1004-4965.2021.007. [27] 祝赢, 柳艳香, 程兴宏, 等. 线性滚动极值处理方法对数值模拟风速的订正研究[J]. 热带气象学报, 2013, 29(4): 681-686. [28] 李照荣, 达选芳, 王有生, 等. 复杂地形条件下风电场预报风速的卡尔曼滤波订正[J]. 高原气象, 2018, 37(5): 1 402-1 412. [29] LI L L, CEN Z Y, TSENG M L, et al. Improving short-term wind power prediction using hybrid improved cuckoo search arithmetic - Support vector regression machine[J]. Journal of Cleaner Production, 2021, 279: 123739. [30] ZHENG Y, GE Y, MUHSEN S, et al. New ridge regression, artificial neural networks and support vector machine for wind speed prediction [J]. Advances in Engineering Software, 2023, 179: 103426. [31] ASADI M, POURHOSSEIN K, IVATLOO B M. GIS-assisted modeling of wind farm site selection based on support vector regression[J]. Journal of Cleaner Production, 2023, 390: 135993. [32] BIANCO L, DJALALOVA I V, WILCZAK J M et al. A Wind Energy Ramp Tool and Metric for Measuring the Skill of Numerical Weather Prediction Models. [J]. Wea Forecasting, 2016, 31(4): 1 137-1 156. -
点击查看大图
图(9) / 表(4)
计量
- 文章访问数: 59
- HTML全文浏览量: 13
- PDF下载量: 15
- 被引次数: 0
粤公网安备 4401069904700003号