面向新能源消纳能力评估的年负荷序列建模及场景生成方法

全年负荷序列是开展中国省级电网新能源消纳能力评估的基础,文中提出了基于聚类分析和马尔可夫链技术的年负荷序列建模和场景生成方法。首先,通过自组织映射对历史负荷数据进行典型日聚类分析,并采用离散马尔可夫链描述不同典型日之间的状态转移特性。针对每类典型日,采用核密度估计和t-Copula函数构建日负荷特性指标的联合概率分布模型。然后,通过马尔可夫链蒙特卡洛随机抽样生成每日的典型日状态和日负荷特性指标。最后,通过构建日负荷序列优化模型,实现每日负荷序列的优化重构,直至生成全年负荷序列场景。算例基于中国某省级电网全年负荷数据进行测试,并利用所生成的负荷序列场景开展未来年度新能源消纳能力的评估,结果验证了所提方法的有效性及实用性。

Modeling and Scenario Generation Method of Annual Load Series for Evaluation of Renewable Energy Accommodation Capacity

Annual load series is the basis of evaluation of renewable energy accommodation capacity in provincial power grid of China. In this paper, modeling and scenario generation methods of annual load series are proposed based on the cluster analysis and Markov chain technology. First, the self-organizing map technology is used on the historical load data for cluster analysis of typical days, and the discrete Markov chain is adopted to describe the state transition characteristics between different typical days. For each type of typical days, the kernel density estimation and t-Copula function are utilized to construct the joint probability distribution model of daily load characteristics. Then, the indices of the typical daily states and daily load characteristics are generated by Markov chain and Monte Carlo random sampling. Finally, through the construction of optimization model for daily load series, the optimization reconstruction of daily load series is realized until the annual load series scenario is generated. The case studies are conducted based on annual load data of a provincial power grid in China. The scenarios of load series generated by the proposed method are also used to evaluate renewable energy accommodation in the next year. The testing results verify the effectiveness and practicality of the proposed method.