| 引用本文: | 杨景旭,郑楷洪,周尚礼,曾璐琨.基于同态加密和K-means聚类算法的用户充电模式聚类和需求响应潜力评估[J].电力自动化设备,2025,45(4):101-109,117 |
| YANG Jingxu,ZHENG Kaihong,ZHOU Shangli,ZENG Lukun.Clustering of user charging patterns and assessment of demand response potential based on homomorphic encryption and K-means clustering algorithm[J].Electric Power Automation Equipment,2025,45(4):101-109,117 |
|
| 摘要: |
| 为解决利用单充电站数据进行用户充电模式提取不准确、不全面的问题，提出在保证用户隐私安全的前提下充分利用区域内多个充电站充电数据来提取用户的充电模式，基于同态加密和K-means聚类算法提出用户充电模式聚类模型和需求响应潜力评估方法。综合考虑不同充电模式在起始充电时间、充电时长和充电功率方面的差异，提出充电模式综合误差作为新的充电模式聚类标准，基于此提出基于手肘法的最优聚类数确定方法。提出基于同态加密算法的用户充电模式提取方案，阐述了方案的参与主体、密钥和随机数管理、数据链式加密操作、算法步骤。提出综合考虑用户日充电频率、充电模式的需求响应时段重合度、充电功率以及充电概率的用户需求响应潜力评估和排序方法，基于此提出充电站充电负荷需求响应潜力计算方法。通过算例验证了所提方法的有效性。 |
| 关键词: 电动汽车 同态加密 充电模式 需求响应 充电站 聚类 K-means聚类算法 |
| DOI：10.16081/j.epae.202412037 |
| 分类号:TM715;U469.72 |
| 基金项目:南方电网数字电网集团有限公司创新项目(210000KC24010003) |
|
| Clustering of user charging patterns and assessment of demand response potential based on homomorphic encryption and K-means clustering algorithm |
|
YANG Jingxu, ZHENG Kaihong, ZHOU Shangli, ZENG Lukun
|
|
China Southern Power Grid Digital Group Co.,Ltd.,Guangzhou 510300, China
|
| Abstract: |
| In order to solve the problem of inaccurate and incomplete user charging pattern extraction using single charging station data, it is proposed to make full use of charging data from multiple charging stations in the region to extract user charging patterns under the premise of guaranteeing the security of userprivacy. And a clustering model for user charging patterns and a method for evaluating the demand response potential based on homomorphic encryption and the K-means algorithm are proposed. Considering the differences between different charging modes in terms of starting charging time, charging duration and charging power, the comprehensive error of charging modes is proposed as a new charging mode clustering criterion. Based on this, an optimal clustering number determination method based on elbow method is proposed. A user charging mode extraction scheme based on homomorphic encryption algorithm is proposed, and the participating subjects, key and random number management, data chaining encryption operation and algorithmic steps of the scheme are described. Then, a evaluating and ranking method for user demand response potential is proposed considering the user’s daily charging frequency, together with the charging pattern demand response time period overlap, charging power and charging probability. Based on this, a method for calculating the charging load demand response potential of charging stations is proposed. The effectiveness of the proposed method is verified through examples. |
| Key words: electric vehicles homomorphic encryption charging pattern demand response charging station clustering K-means clustering algorithm |
