基于条件Copula函数的风电功率区间预测

Interval Prediction for Wind Power Based on Conditional Copula Function

提出一种基于多个关联因素的联合分布函数的风电功率的短期/超短期区间预测方法,以已知的影响因素为条件,建立待预测风电功率的条件Copula联合分布函数。并基于此,计算满足给定置信度水平下,待预测风电功率所在区间的上、下界。该方法充分利用Copula函数可以较为准确反映多个因素之间关联概率分布函数的优点,借助待预测量与影响因素之间的相依关系进行预测,拟提高区间预测的精度。介绍了区间预测的定义和基本概念,提出了连接多个因素的条件Copula函数的构造方法,并基于此提出了区间预测的实现过程和方法。最后,将所提方法应用于爱尔兰国家下辖的科克州的风电功率预测,并研究和分析了多组参数设置下的预测结果,验证所提方法的有效性。

This paper proposed a novel short-time/very short time interval prediction method for wind power, based on the joint distribution function of its influence factors. A conditional joint distribution function based on copula function was constructed, which took the known wind power influence factors as conditions. After obtaining the copula function, the upper and lower bound of given confidence interval which the wind power lying in could be calculated. The proposed method takes full advantages of copula function that can well reflect the relationships between the influence factors and forecast wind power. By the aid of copula function, the prediction accuracy could be improved. Firstly, the basic concept of interval prediction was introduced and analyzed. Then, the construction method of conditional copula function was proposed. In the basis of copula function, the wind power interval prediction method was given. At last, the proposed method was applied to the wind power prediction in Cork, Irish national wind farm, by setting various parameters of the method. The application results verify the effectiveness of the method. © 2016 Chin. Soc. for Elec. Eng.