光伏发电出力的条件预测误差概率分布估计方法

光伏发电出力的可预测性较低,相比点预测而言,光伏发电出力的概率性预测能够提供更多的信息,有利于电力系统的安全经济运行。提出了一种基于Copula理论的光伏发电出力的条件预测误差分布估计方法。采用Copula函数对光伏实际出力与点预测的联合概率分布进行建模,实现了任意点预测对应的光伏实际出力的条件概率分布的估计。针对天气状况,对光伏预测精度影响较大的实际情况,采用聚类的方法按天气类型将历史数据进行分类,针对每类天气类型的光伏预测误差分别进行建模以提高预测误差估计的准确度。以2014全球能源预测竞赛(GEFC 2014)中的光伏出力数据进行了实证分析,验证了所提出方法对光伏出力条件预测误差估计的有效性,结果表明提出的方法在校准性和锐度方面均优于常用的正态分布的预测误差估计方法。

A Method of Probabilistic Distribution Estimation of Conditional Forecast Error for Photovoltaic Power Generation

Owing to the poor predictability of photovoltaic power, its probabilistic forecast provides more information about the underlying uncertainties compared with the traditional point forecast. This paper proposes a Copula theory based method to estimate the conditional forecasting error in photovoltaic power generation. The joint probability between the actual power output and its forecast is modeled using the Copula function. The conditional forecast error corresponding to each photovoltaic forecast level is then derived from this joint probability model. Considering the fact that weather has a strong impact on the accuracy of photovoltaic forecasting, cluster technique is used to divide the data according to weather types. A joint distribution model each is developed for the respective weather types to estimate the forecast error. Empirical study is carried out to validate the proposed model using the data from Global Energy Forecasting Competition 2014. The results show that the proposed method has improved both the calibration and sharpness of photovoltaic generation probabilistic forecast compared with the traditional normal-distribution-based probabilistic forecast method. This work is supported by National Natural Science Foundation of China (No. 51307092, No. 51325702) and State Grid Corporation of China.