中文摘要：

提出一种光伏功率预测方法,该方法利用相似变量的基值点误差,采用组合权重法求取相似误差,确定相似日,然后将相似日的输出功率按不同权重系数加权得到预测功率。根据最小鉴别信息原理,将主观权重和客观熵权有效融合获得相似日选取的组合权重系数,功率权重系数按相似性指标生成。某光伏电站实测数据算例仿真表明,所提方法可选出相似度高的相似日,提高了光伏输出功率的预测精度。

英文摘要：

A method of photovoltaic power output forecasting is proposed,which,based on the deviations of similar variable at basic points,applies the combined weight to calculate the similarity deviation for the similar day selection,and then forecasts the power output by calculating the weighted sum of the similar day power outputs. The combined weight coefficient of selected similar day is the combination of the subjective weight and objective entropy according to the minimum discrimination information principle while the power weight coefficient is generated according to the similarity index. A case study based on the actual data of a photovohaic plant shows that,the similar days selected by the proposed method have higher similarity, resulting in higher power forecasting accuracy.