考虑预测误差时序分布特性的含风电机组组合模型

发布时间：2018-08-16 13:28

【摘要】：风功率预测误差的高精度拟合与置信水平的科学选取是保障风电高效参与日前机组组合(UC)的关键问题,该文将在同时考虑上述两方面问题基础上,研究含风电的UC决策模型。首先,基于误差特性分析提出一种时序下的预测误差分段拟合方法,利用t location-scale分布进行拟合,以改善厚尾效应,提高拟合精度,更可在时序上与UC决策相配合;其次,构建了可以同时考虑传统成本、额外备用成本与风险成本的双分位点型UC决策模型,通过不同成本间的制约关系平衡置信水平的选取,通过不同置信水平的划分指导备用分类,通过时变置信水平适应误差时序分段分布,以此使模型更具经济性、针对性与适用性;最后,采用带有启发式搜索原则的改进混合粒子群算法,求解文中的多变量混合整数规划模型,算例结果验证了所提方法的有效性。
[Abstract]:The high precision fitting of wind power prediction error and the scientific selection of confidence level are the key problems to ensure the wind power to participate in the (UC) of pre-day units efficiently. This paper will study the UC decision model with wind power on the basis of considering the above two problems at the same time. First of all, based on the analysis of error characteristics, a prediction error subsection fitting method based on time series is proposed, which is fitted by t location-scale distribution to improve the effect of thick tail and improve the fitting accuracy, which is more suitable for UC decision making in time series. A two-point UC decision model which can consider the traditional cost, extra reserve cost and risk cost at the same time is constructed. Through the selection of the balance confidence level between different cost constraints, the standby classification is guided by the division of different confidence levels. The time-varying confidence level adapts to the time series distribution of error to make the model more economical, targeted and applicable. Finally, an improved hybrid particle swarm optimization algorithm with heuristic search principle is adopted. The multivariable mixed integer programming model is solved, and the effectiveness of the proposed method is verified by an example.
【作者单位】：电网智能化调度与控制教育部重点实验室(山东大学);电力系统及发电设备控制和仿真国家重点实验室(清华大学);
【基金】：国家重点基础研究发展计划项目(973项目)(2013CB228205) 国家自然科学基金项目(51177091,51477091) 山东省青年科学家奖励基金项目(BS2015NJ005) 山东大学基本科研业务费资助项目(2015GN001)~~
【分类号】：TM614

【相似文献】