| 引用本文: | 谷紫文,李 鹏,郎 恂,等.基于变分模态分解和密度峰值快速搜索的电力负荷曲线可控聚类模型[J].电力系统保护与控制,2021,49(8):118-127.[点击复制] |
| GU Ziwen,LI Peng,LANG Xun,et al.A controllable clustering model of the electrical load curve based on variational modedecomposition and fast search of the density peak[J].Power System Protection and Control,2021,49(8):118-127[点击复制] |
|
| 摘要: |
| 电力负荷曲线作为一种非平稳信号，可以看作由宽平稳的低频分量和非平稳的高频分量构成。针对负荷数据的时间多粒度构成特点，提出了一种基于变分模态分解和密度峰值快速搜索的负荷可控聚类模型。原始负荷曲线通过变分模态分解算法被分解为低频，中频和高频三个模态分量。首先，利用负荷曲线的低频模态分量实现簇间的时间粗粒度聚类。然后，在子类中添加中频分量实现簇内的时间细粒度聚类。使用OpenEI数据集对所提模型进行了有效性验证，并与不同聚类算法对原始负荷数据直接聚类进行对比。实验结果表明该模型可以实现不同时间颗粒度的合理聚类。 |
| 关键词: 负荷曲线聚类 变分模态分解 密度峰值聚类 智能电网 数据驱动 |
| DOI：DOI: 10.19783/j.cnki.pspc.200713 |
| 投稿时间：2020-06-23修订日期：2020-08-17 |
| 基金项目:国家自然科学基金项目资助(61763049)；云南省应用基础研究重点课题(2018FA032) |
|
| A controllable clustering model of the electrical load curve based on variational modedecomposition and fast search of the density peak |
| GU Ziwen1,LI Peng1,LANG Xun1,YU Yixuan1,SHEN Xin2,CAO Min2 |
| (1. School of Information Science and Engineering, Yunnan University, Kunming 650091, China;
2. Electric Power Research Institute, Yunnan Power Grid Co., Ltd., Kunming 650217, China) |
| Abstract: |
| As a kind of non-stationary signal, the electrical load curve can be regarded as composed of a wide-stationary low-frequency component and a non-stationary high-frequency component. Aiming at the temporal multi-granularity characteristics of load data, a controllable clustering model of the electrical load curve based on variational mode decomposition and fast search of the density peak is proposed. The original load curve is decomposed into three modal components of low-, intermediate-, and high-frequency by variational mode decomposition. First, the low-frequency modal components of the load curve are used to achieve coarse-grained clustering among clusters. Then, the intermediate-frequency components are superimposed on the subclasses to achieve fine-grained clustering within the cluster. The validity of the proposed model is verified using the OpenEI data set and compared with different clustering algorithms for the direct clustering of the original load data. The experimental results show that the model can achieve reasonable clustering at different time granularities.
This work is supported by the National Natural Science Foundation of China (No. 61763049) and the Key Project of Applied Basic Research of Yunnan Province (No. 2018FA032). |
| Key words: load curve clustering variational mode decomposition density peak clustering smart grid data driven |
