采用横纵向误差平移插值修正的风电预测精度改善方法  

A Forecasting Accuracy Improvement Method for Wind Power Based on Phase and Level Errors Translating and Interpolating Correction

作　　者：胡亚伟[1] 王筱[2] 晁勤 蔺红[1] 陈哲[4] 王一波[1] 

机构地区：[1]新疆大学电气工程学院,新疆维吾尔自治区乌鲁木齐市830047 [2]国网新疆电力公司,新疆维吾尔自治区乌鲁木齐市830002 [3]克拉玛依理工学院筹建办公室,新疆维吾尔自治区克拉玛依市834000 [4]奥尔堡大学工程与科学学院

出　　处：《电网技术》2015年第10期2758-2765,共8页Power System Technology

基　　金：国家自然科学基金项目(51267020);国家国际科技合作专项资助(2013DFG61520);高等学校博士学科点专项科研基金博导类联合资助项目(20126501110003)~~

摘　　要：目前风电预测功率精度偏低,评估指标常用纵向误差,很少针对横向误差(延迟或超前时间)进行统计。提出了基于横、纵向误差平移修正的风电预测精度改善方法,给出用于风电功率预测的不变趋势功率持续时间的概念,并将风电场日前功率预测曲线和实际功率曲线划分为不变趋势功率增加段、降低段和保持段3种模式。建立了各不变趋势持续时间的计算模型,采用概率统计法获得预测时间值与实际时间值之间的偏差值及滞后或超前方向,利用平移法和插值法联合横纵向误差修正日前功率预测误差。应用上述方法对新疆某地区风电功率预测曲线进行仿真计算,结果如下:该地区风电功率预测横向误差为3.78 h,方向为超前,纵向绝对误差值为40.05 MW;采用平移插值修正后,横向误差值为2.56 h,降低了32.34%,纵向误差值为32.58 MW,降低了18.65%。从而表明该方法可以有效提高风电短期预测功率精度。At present, accuracy of short-term wind power forecasting is low, level errors are widely used as evaluation index, while few statistics about phase errors （delay or ahead of time） are studied. In this paper, a method based on translating and interpolating correction of phase and level errors to improve accuracy of wind power forecasting is put forward, concept of phase error constant trend duration is proposed, and day-ahead wind power forecasting and actual curve is classified into three modes, i.e. constant trend power of increasing, decreasing and unchanging. Constant trend duration computational model is established based on probabilistic and statistical methods to obtain deviation and lagging or leading direction between forecasting and actual time value, using translated and interpolated method and combining phase and level errors to correct day-ahead power forecasting error. Results show that with wind power forecasting curve simulation for three months in a region of Xinjiang, forecasting phase error of the region is 3.78 h in leading direction, and absolute level error is 40.05 MW. After translated and interpolated correction, level error is 2.56 h, reduced by 32.34%, level absolute error is 32.58 MW, reduced by 18.65%. The accuracy of short-term wind power forecasting is effectively improved.

关 键 词：风电功率预测 横向误差 纵向误差 概率统计 平移插值修正 

分 类 号：TM614[电气工程—电力系统及自动化]