高比例新能源已成为现代电力系统的突出特征,新能源发电功率的间歇性、波动性给电力系统带来显著不确定性,极大增加电力系统安全和稳定运行的风险。概率预测可实现新能源电力系统预测不确定性的有效量化,相较于经典确定性预测可提供更加全面的预测信息,为电力系统分析与决策提供关键数据支撑。该文首先从数学角度阐述预测的基本定义,进而分析预测不确定性的产生机理,阐述概率预测的时间尺度、评价指标等基本概念;其次从概率预测的本质科学问题及其数学表达的角度阐释概率预测的数学原理;然后按照不同分类方式综述概率预测的常用方法;最后总结概率预测研究中的问题与挑战。

High proportion renewable energy has become the prominent feature of modern power system. The intermittency and volatility of renewable energy generation bring significant uncertainties for the power systems, and thus increase the risks to power system security and stability. Probabilistic forecasting can effectively realize the quantification of prediction uncertainty in power systems with renewable energy sources. Compared with classical deterministic forecasting, probabilistic forecasting can provide more comprehensive forecasting information, and key data supports for the analysis and decision-making of power systems. This paper mathematically elaborated the basic definitions of forecasting and further analyzed the generating mechanism of forecasting uncertainties. Basic concepts including forecasting horizons and evaluation indices for probabilistic forecasting were expounded detailedly. Then, mathematical principles for probabilistic forecasting were demonstrated from the point of essential scientific problems and mathematical expressions. The commonly used methodologies for probabilistic forecasting were investigated from different types of classifications. Finally, the problems and challenges for researches on probabilistic forecasting were summarized in detail.