上海交通大学学报

上海交通大学学报

• •    

考虑源荷不确定性的信息间隙决策理论-谱聚类典型场景生成方法（网络首发）

  

  1. 1. 长沙理工大学防灾减灾全国重点实验室；2. 国网湖南省电力有限公司经济技术研究院；3. 国网长沙供电公司
  • 基金资助:
    国家自然科学基金（52277076； 52307081）； 国网湖南省电力有限公司科技项目（5216A221001L）

Information Gap Decision Theory-Spectrum Clustering Typical Scenario Generation Method Considering Source Load Uncertainty

  1. （1. National Key Laboratory of Disaster Prevention and Mitigation, Changsha University of Science and Technology, Changsha 410114, China；2. State Grid Hunan Electric Power Company Limited Economic & Technical Research Institute, Changsha 410004, China；3. State Grid Changsha Power Supply Company, Changsha 410015, China）

PDF (PC)

14

摘要： 高比例新能源和动态负荷的接入给电力系统带来显著的源荷双向不确定性。强不确定性使得调度规划面临高维决策空间，增加了规划风险。因此，提出一种考虑源荷不确定性的信息间隙决策理论（Information gap decision theory， IGDT）-谱聚类典型场景生成方法，为多源联合系统运行方式的确定提供更精确的规划场景。首先，利用IGDT理论无需考虑不确定量概率分布的优点，对源荷不确定性进行有效量化。综合IGDT鲁棒和IGDT机会模型刻画源荷波动范围，并采用拉丁超立方抽样法生成代表各不确定性情形的原始场景，从而保证样本空间的充分性与准确性。其次，针对源荷不确定性导致的原始场景规模庞大问题，引入考虑系统调节能力的谱聚类方法，挖掘不同源荷波动对调度决策具有重要影响的特征向量，从而将复杂原始场景集缩减为具有代表性的源荷典型场景。最后，通过某省电网实际系统及运行数据仿真分析表明，所提方法在考虑了源荷双向不确定性后，相较于传统谱聚类方法，多生成4种典型场景，且综合平均皮尔逊相关系数提高了8.76%，综合平均欧氏距离缩减了43.48%，聚类场景与实际场景更相似。

关键词: 源荷不确定性, 信息间隙决策理论, 谱聚类, 典型场景生成, 调度规划

Abstract: The high proportion of new energy and dynamic load bring significant bidirectional uncertainty of source and load to power system. Strong uncertainty makes scheduling planning face high dimensional decision space and increases planning risk. Therefore, an information gap decision theory (IGDT) -spectral clustering typical scenario generation method considering the uncertainty of source load is proposed to provide a more accurate planning scenario for the determination of the operation mode of multi-source joint systems. Firstly, the source load uncertainty can be quantified effectively by using IGDT theory without considering the advantage of uncertain quantitative probability distribution. The source load fluctuation range is described by IGDT robust model and IGDT chance model, and the original scenario representing each uncertainty situation is generated by Latin hypercube sampling method, so as to ensure the adequacy and accuracy of sample space. Secondly, in view of the huge scale of the original scenario caused by the uncertainty of the source load, a spectral clustering method considering the adjustment ability of the system is introduced to mine the feature vectors of different source load fluctuations that have an important impact on the scheduling decision, so as to reduce the complex original scenario set to a representative typical scenario of the source load. Finally, through the simulation analysis of the actual system and operation data of a provincial power grid, it is shown that compared with the traditional spectral clustering method, the proposed method generates 4 more typical scenarios after considering the bidirectional uncertainty of source load, and the comprehensive average Pearson correlation coefficient is increased by 8.76%, the comprehensive average Euclidian distance is reduced by 43.48%, and the clustering scenario is more similar to the actual scenario.

Key words: source load uncertainty, information gap decision theory (IGDT), spectral clustering, typical scenario generation, dispatch planning

中图分类号: