#### 期刊菜单

Classification of Electric Power Load in Xinjiang Based on Daily Load Curve
DOI: 10.12677/SG.2019.96032, PDF, HTML, XML, 下载: 454  浏览: 729

Abstract: Power load classification is the basic data and key indicators of power system planning and operation control. The location characteristics and load characteristics of Xinjiang region obviously determine the classification of power load is particularly important. Therefore, it is of great engineering value to carry out accurate classification research of power load in Xinjiang. This paper proposes a power load classification method based on daily load curve and K-means clustering algorithm in Xinjiang. Firstly, the load characteristics of Xinjiang region are analyzed, and the load classification ideas and classification indicators of Xinjiang region using daily load curve data processing are proposed. Then the load classification in Xinjiang is realized based on K-means clustering. The results show the correctness and effectiveness of the proposed method.

1. 引言

2. 新疆地区负荷曲线的分离

2.1. 新疆负荷特点

2.2. 新疆负荷聚类指标的确定

Table 1. The connotation of clustering indicators

2.3. 新疆地区日负荷曲线分离

Table 2. Daily load curve separation results

3. 基于K-Means的新疆负荷聚类

3.1. K-Means聚类算法

K-means聚类算法，也被称为K-均值聚类算法，与其它聚类算法相比，该算法聚类效果良好。K-means聚类算法的过程如图1所示。

3.2. 基于K-Means聚类算法的新疆地区典型区域负荷分类的实现

K-means聚类算法聚类数的确定有手肘法和轮廓系数法。手肘法主要借助SSE指标，SSE计算的是所有样本点的聚类误差，通过SSE可以从某种程度上反应聚类效果。SSE的计算如式(1)所示。

(1)

Figure 1. K-means clustering algorithm flow chart

Figure 2. SSE calculation result

Table 3. Partial site clustering indicator (Xia Da)

Table 4. Clustering results (summer)

Table 5. Clustering results (winter)

Figure 3. Clustering results (summer)

Figure 4. Clustering results (winter)

3.3. 聚类效果评价

MIA、CDI、SMI、DBI、SI这5个指标在聚类效果评价中较为常用，本文取MIA和CDI这两个指标进行分析。Mean Index Adeqquacy (MIA)表示的是分配给聚类的每个输入向量与其中心之间的平均距离，MIA指标越小，聚类结果内部越紧凑，其值与聚类效果成负相关。Clustering Dispersion Indicator (CDI)，表示的是同一集群中输入向量之间的平均基础距离与类代表负荷曲线之间的基础距离的比率，其值与聚类效果也成负相关。MIA和CDI的计算结果如表6所示。

Table 6. MIA and CDI calculation results

4. 结论

 [1] 贺仁睦, 周文. 电力系统负荷模型的分类与综合[J]. 电力系统自动化, 1999, 23(19): 12-16. [2] 张红斌. 电力系统负荷模型结构与参数辨识的研究[D]: [博士学位论文]. 北京: 华北电力大学, 2003. [3] 鞠平, 陈谦, 熊传平, 等. 基于日负荷曲线的负荷分类和综合建模[J]. 电力系统自动化, 2006, 30(16): 6-9. [4] 陈烨, 吴浩, 史俊祎, 等. 奇异值分解方法在日负荷曲线降维聚类分析中的应用[J]. 电力系统自动化, 2018, 42(3): 105-111. [5] Jain, A.K. (2010) Data Clustering: 50 Years beyond K-Means. Pattern Recognition Letters, 31, 651-666. https://doi.org/10.1016/j.patrec.2009.09.011 [6] Rahman, M.A. and Islam, M.Z. (2014) A Hybrid Clustering Technique Combining a Novel Genetic Algorithm with K-Means. Knowledge-Based Systems, 71, 345-365. https://doi.org/10.1016/j.knosys.2014.08.011 [7] 徐邦恩, 蔺红. 基于改进模糊聚类的典型日负荷曲线选取方法[J]. 电测与仪表, 2019, 56(4): 21-26. [8] Zhou, L., Peng, Z.W., Deng, C.R., Qi, X.Z. and Li, P.Q. (2018) A Generalized Synthesis Load Model Considering Network Parameters and Allvanadium Redox Flow Battery. Protection and Control of Modern Power Systems, 3, 315-327. https://doi.org/10.1186/s41601-018-0105-1