设为首页
加入收藏
Comparison of Short-term Forecast Method of Wind Power in Wind Farm
-
摘要: 通过开展湖北省九宫山风电场短期风电功率预报方法的研究，以不断提高预报准确率，为风电场提供更有价值的预报服务，该文利用MM5耦合CALMET模式模拟风电场风速资料，采用物理法和动力统计法探讨风电场各种情况下预报应用效果。结果表明：模拟风速释用订正能有效降低风速预报误差，但难以修正预报趋势；动力统计法更适用于九宫山风电场的复杂山区地形，可能由于该方法能自发适应风电场地理位置；采用实测数据建立的风电功率预报模型优于理论风电功率模型，这也与风机实际运行环境会影响风机输出功率有关。Abstract: To further improve the accuracy of wind energy forecast and provide more valuable service, several wind energy forecast methods are studied comparatively in Jiugongshan Wind Farm. Based on the wind speed simulation results by the aggregative model CALMET coupled with MM5 (the model resolution is 200 m), the principle method and dynamic-statistical method are used to discuss 24-h forecast effect, with the temporal resolution set to 15 minutes. There are three kinds of principle method to discuss the effect of simulated wind speed correcting and the wind energy forecast model based on observations. The dynamic-statistical method forecast by establishing a rolling model using the simulated data of last period every day.The fine-scale simulation can obviously forecast the variation trend of wind speed. The correcting of simulated wind speed can effectively reduce the wind speed error and improve the forecasting accuracy, but it is difficult to revise the changing trend of simulated wind speed.The dynamic-statistical method is much more suitable for the complex topography mountainous terrain, and the monthly relative mean square root error is 14%—26% from July to December in 2011, which might be the result of its spontaneous adaption for terrain conditions.The wind energy forecast model based on observations is better than those based on theoretical model and can effectively reduce the forecasting error, because the wind farm environment has unique effects on the output power of fan.Furthermore, it is discovered that the wind energy forecast in southern mountain area is much more difficult than in north area. The fine-scale simulation should be used to reduce the infection of terrain; the method of simulated wind speed correcting must consider the different situation of the wind farm; the extreme weather events must be considered, and effects of different weather especially meteorological disaster such as ice-coating and thunderstorm should be deeply studied. These results enhance the service effect at Jiugongshan Wind Farm in Hubei Province, and more research should be carried out to improve the forecast accuracy.
-
图 1 风电功率短期预报物理法流程图
Fig. 1 The flow chart of wind energy short-term forecast using principle method
图 2 实际风电功率预报模型拟合曲线
Fig. 2 The fitting curve of wind energy forecast model using observations
图 3 2011年7—12月九宫山风电场风电功率预报相对均方根误差比较
Fig. 3 The relative mean square root error of wind energy forecast at Jiugongshan Wind Farm from July to December in 2011
表 1 模拟风速订正模型
Table 1 The correcting model of simulated wind speed
订正条件 订正模型 υ模拟 < 3 m·s-1 y=x(不订正) υ模拟≥3 m·s-1 y=-0.0015x4+0.0544x3-0.6438x2+3.5233x-2.201 注：x为模拟风速，单位：m·s-1; y为订正风速，单位：m·s-1。 
下载: 导出CSV
表 2 模拟风速订正效果检验
Table 2 The correcting effect test of simulated wind speed
月份 条件 相关系数 均方根误差/(m·s-1) 平均绝对误差/(m·s-1) 7 实测与模拟 0.584 3.22 2.5 实测与订正 0.583 2.90 2.3 8 实测与模拟 0.653 3.00 2.4 实测与订正 0.633 2.85 2.3 9 实测与模拟 0.365 3.77 3.0 实测与订正 0.381 3.12 2.5 10 实测与模拟 0.230 4.01 3.2 实测与订正 0.207 3.70 3.0 11 实测与模拟 0.496 3.24 2.9 实测与订正 0.471 2.95 2.6 12 实测与模拟 0.412 2.38 1.8 实测与订正 0.396 2.38 1.8
下载: 导出CSV
表 3 风电功率预报模型
Table 3 The wind energy forecast model
项目 风速分段/(m·s-1) 风电功率预报方程 x < 3 W=0 理论风电功率预报模型 3≤x < 12 W=-3.5293x3+90.842x2-626.38x+1385.9 x≥12 W=850 x < 3 W=0 实际风电功率预报模型 3≤x < 13 W=-0.0621x4+0.7055x3+11.477x2-80.801x+148.97 x≥13 W=850 注：x为预报风速，单位：m·s-1；W为风电功率，单位：kW。
下载: 导出CSV
表 4 风电功率预报效果检验比较
Table 4 The comparison of wind energy forecast by different methods
月份 方法 相关系数 相对均方根误差/% 平均绝对误差/kW 实测平均功率/kW 预报平均功率/kW 7 物理法1 0.593 30 2760.5 3579.4 4195.2 物理法2 0.605 27 2385.7 3151.3 物理法3 0.604 26 2344.2 2837.0 动力统计法 0.644 24 2409.2 2984.1 8 物理法1 0.722 25 2427.4 4033.9 4045.1 物理法2 0.706 25 2316.2 3100.8 物理法3 0.699 25 2305.9 2859.9 动力统计法 0.624 26 2703.2 3091.9 9 物理法1 0.436 33 2938.6 1368.7 3604.4 物理法2 0.500 23 2028.7 2597.5 物理法3 0.504 19 1783.4 2323.5 动力统计法 0.414 18 1974.7 2731.1 10 物理法1 0.217 34 3007.6 2220.3 2501.6 物理法2 0.228 30 2646.7 2091.3 物理法3 0.227 28 2467.5 1890.8 动力统计法 0.154 25 2492.6 2287.4 11 物理法1 0.480 37 3875.6 3781.2 648.4 物理法2 0.473 36 3702.8 1030.0 物理法3 0.558 33 3152.9 774.1 动力统计法 0.683 24 2501.2 2205.9 12 物理法1 0.319 23 1817.3 1512.0 649.6 物理法2 0.289 22 1796.5 1004.0 物理法3 0.408 18 1291.9 573.6 动力统计法 0.211 14 1417.3 1718.6
下载: 导出CSV
-
[1] 李俊峰.风光无限——中国风电发展报告2011.北京:中国环境科学出版社, 2011. [2] 李俊峰.2012中国风电发展报告.北京:中国环境科学出版社, 2012. [3] 陈正洪, 许杨, 许沛华, 等.风电功率预测预报技术原理及其业务系统.北京:气象出版社, 2013. [4] 罗如意, 林晔, 钱野.世界风电产业发展综述.可再生能源, 2010, 28(2): 14-17. http://www.cnki.com.cn/Article/CJFDTOTAL-NCNY201002005.htm [5] 王勇, 李照荣, 李晓霞, 等.风电功率预报方法研究进展.干旱气象, 2011, 29(2): 156-160. http://www.cnki.com.cn/Article/CJFDTOTAL-GSQX201102006.htm [6] 孙川永. 风电场风电功率短期预报技术研究. 兰州: 兰州大学, 2009. [7] 彭晖. 风电场风电量短期预测技术研究. 南京: 东南大学, 2009. [8] 韩爽. 风电场功率短期预测方法研究. 北京: 华北电力大学, 2008. [9] 李杏培. 风电场风速及风电机组发电量的短期预报方法研究. 北京: 华北电力大学, 2009: 25-28. [10] 孙川永, 陶树旺, 罗勇, 等.高分辨率中尺度数值模式在风电场风速预报中的应用.太阳能学报, 2009, 30(8): 1097-1099. http://www.cnki.com.cn/Article/CJFDTOTAL-TYLX200908021.htm [11] 穆海振, 徐家良, 柯晓新, 等.高分辨率数值模式在风能资源评估中的应用初探.应用气象学报, 2006, 17(2): 152-159. doi: 10.11898/1001-7313.20060204 [12] 张小伟, 李正泉, 杨忠恩, 等.多作业管理方式在风能资源数值模拟中的应用.应用气象学报, 2010, 21(6): 747-753. doi: 10.11898/1001-7313.20100612 [13] 王建捷, 胡欣. MM5模式中不同对流参数化方案的比较试验.应用气象学报, 2001, 12(1): 41-53. http://qikan.camscma.cn/jams/ch/reader/view_abstract.aspx?file_no=20010105&flag=1 [14] 朱蓉, 徐大海.中尺度数值模拟中的边界层多尺度湍流参数化方案.应用气象学报, 2004, 15(5): 543-555. http://qikan.camscma.cn/jams/ch/reader/view_abstract.aspx?file_no=20040567&flag=1 [15] 黄嘉佑.气象统计分析与预报方法.北京:气象出版社, 2004. [16] 国家能源局. 风电场功率预测预报管理暂行办法. 2011. [17] 张礼达, 张彦南.气象灾害对风电场的影响分析.电力科学与工程, 2009, 25(11): 28-30. doi: 10.3969/j.issn.1672-0792.2009.11.007 -
计量
- 摘要浏览量: 2991
- HTML全文浏览量: 997
- PDF下载量: 1641
- 被引次数: 0
