| 引用本文: | 林 俐,潘险险.基于分裂层次半监督谱聚类算法的风电场机群划分方法[J].电力自动化设备,2015,35(2): |
| LIN Li,PAN Xianxian.Wind turbine grouping based on semi-supervised split-hierarchical spectral clustering algorithm for wind farm[J].Electric Power Automation Equipment,2015,35(2): |
|
| 摘要: |
| 在谱聚类技术的基础上，面向风电场动态等值建模，提出一种基于分裂层次半监督谱聚类算法的风电场机群划分方法。首先根据风电机组的实测运行数据，构造一个可以体现原始数据结构且能为分类提供更多有效信息的特征向量矩阵Y。进而利用获取的部分样本组的先验信息，采用自顶向下的簇分裂策略，对Y中的样本组进行半监督聚类划分，得到风电机组的机群划分结果，采用容量加权法计算各机群等值风电机组的参数，建立风电场的动态等值模型。以某实际风电场为例进行仿真验证，结果表明，采用所提方法建立的动态等值模型与详细模型较接近，能够较准确地反映风电场的动态响应特性。 |
| 关键词: 分裂层次 半监督 谱聚类 风电场 机群划分 聚类算法 动态等值 |
| DOI： |
| 分类号: |
| 基金项目:国家自然科学基金重大项目（51190103）；国家科技支撑计划项目（2013BAA02B01） |
|
| Wind turbine grouping based on semi-supervised split-hierarchical spectral clustering algorithm for wind farm |
|
LIN Li, PAN Xianxian
|
|
State Key Laboratory of Alternate Electrical Power System with Renewable Energy Sources，North China Electric Power University，Beijing 102206，China
|
| Abstract: |
| A method of wind turbine grouping based on semi-supervised split-hierarchical algorithm is proposed，which builds the dynamically equivalent model of wind farm based on the spectral clustering technology. An eigenvector matrix Y is formed based on the measured operating data of wind turbines，which reflects the original data structure and provides more effective information for clustering. With the available prior information of some samples，the semi-supervised split-hierarchical spectral clustering strategy is adopted to divide the sample groups of Y into several clusters as the wind turbine groups. The capacity weighting method is applied to calculate the parameters of each equivalent wind turbine group and build the dynamically equivalent model of wind farm. As an example，results of the simulation for an actual wind farm show that，the dynamically equivalent model built approximates to the detailed model，accurately reflecting the dynamic response characteristics of wind farm. |
| Key words: split hierarchy semi-supervision spectral clustering wind farms wind turbine grouping clustering algorithms dynamic equivalence |
