| 摘要: |
| 电力负荷曲线聚类是配用电系统的基础，对负荷管理具有重大意义。采用基于核方法的聚类算法提高负荷曲线聚类的准确性，通过点积的方式构造核矩阵，再将数据映射到高维空间中进行聚类，进而加大数据的可分性。同时，针对核矩阵的规模大、计算复杂的问题，提出使用核主成分与缩减矩阵规模对该方法进行优化。实验过程中采用美国能源部开发能源信息网站提供的负荷数据进行聚类，并以Davies-Bouldin聚类有效性指标评估效果。结果表明该方法具有较好的划分能力，可以提高负荷曲线聚类的准确性。 |
| 关键词: 负荷曲线 聚类算法 核矩阵 核主成分分析 削减矩阵 |
| DOI： |
| 分类号: |
| 基金项目:中央高校基本科研业务费专项资金资助项目（12MS121） |
|
| Load curve clustering based on Kernel K-means |
|
ZHAO Wenqing, GONG Yaqiang
|
|
School of Control and Computer Engineering，North China Electric Power University，Baoding 071003，China
|
| Abstract: |
| As the basis of distribution and utilization system，the load curve clustering is of great significance for load management. A clustering algorithm based on the kernel method is proposed to improve the accuracy of the load curve clustering，which applies the dot product to construct the kernel matrix and maps the data into a high-dimensional space to increase the data divisibility for clustering. Aiming at the large scale and calculation complexity of the kernel matrix，the kernel principal component analysis and the kernel matrix size reduction are adopted to optimize the proposed method. As an experiment，the load data provided by the United State Department of Energy Development Energy Information Website are clustered and its effectiveness is assessed by the Davies-Bouldin index，which show that，the proposed method has better classification capability and the accuracy of load curve clustering is improved. |
| Key words: load curve clustering algorithms kernel matrix kernel principal component analysis matrix size reduction |
