利用模糊信息粒化与支持向量机的空间负荷预测方法  

Spatial Load Forecasting Method Using Fuzzy Information Granulation and Support Vector Machine

作　　者：肖白[1] 赵晓宁 姜卓[2] 施永刚 焦明曦 王徭 XIAO Bai;ZHAO Xiaoning;JIANG Zhuo;SHI Yonggang;JIAO Mingxi;WANG Yao(School of Electrical Engineering,Northeast Electric Power University,Jilin 132012,Jilin Province,China;School of Computer Science and Technology,Beihua University,Jilin 132021,Jilin Province,China;Tonghua Power Supply Company,State Grid Jilin Electric Power Company Co.,Ltd.,Tonghua 134001,Jilin Province,China;Changchun Power Supply Company,State Grid Jilin Electric Power Company Co.,Ltd.,Changchun 130021,Jilin Province,China)

机构地区：[1]东北电力大学电气工程学院,吉林省吉林市132012 [2]北华大学计算机科学技术学院,吉林省吉林市132021 [3]国网吉林省电力有限公司通化供电公司,吉林省通化市134001 [4]国网吉林省电力有限公司长春供电公司,吉林省长春市130021

出　　处：《电网技术》2021年第1期251-258,共8页Power System Technology

基　　金：国家自然科学基金项目(51177009);吉林省产业创新专项基金项目(2019C058-7);吉林省教育厅科技项目(JJKH20180442KJ)。

摘　　要：若直接使用实测负荷数据最大值进行空间负荷预测,则元胞负荷中的异常数据会导致预测结果精度降低,考虑到通过确定并利用元胞负荷合理最大值可以明显改善预测精度,提出一种基于模糊信息粒化与支持向量机的空间负荷预测方法。首先构建电力地理信息系统,并在其中生成2类元胞。其次按照时间尺度的长短区分Ⅰ类元胞负荷颗粒度的粗细,通过划分模糊粒化窗口,建立合理的模糊集对Ⅰ类元胞细颗粒度下的历史负荷数据进行模糊信息粒化,进而确定出Ⅰ类元胞粗颗粒度下的历史负荷的合理最大值。然后采用支持向量机模型,对粗颗粒度下的Ⅰ类元胞负荷进行预测。最后确定Ⅰ类元胞负荷密度均衡系数,求取分类负荷密度指标,结合用地信息求得各Ⅱ类元胞负荷预测值,从而实现对空间电力负荷预测结果的网格化。工程实例表明了该方法的实用性和有效性。When the spatial load is forecasted directly with the actually-measured maximum load value, the abnormal data of the cell load will result in the inaccuracy of the forecasting. Considering that the reasonable maximum value of the cell load can be determined and utilized to improve the accuracy of the forecasting, a spatial load forecasting method based on fuzzy information granulation support vector machine is proposed. First, a power geographic information system is established, in which two types of cells are generated. Secondly, according to the length of the time scale, the thickness of the class Ⅰ cell load granularity is distinguished. By classifying the fuzzy granulation windows and establishing reasonable fuzzy sets, the fuzzy information granulation is performed on the historical class Ⅰ cell load data under fine particle size, and then the reasonable maximum value of the historical load data under the coarse granularity in class Ⅰ cell is generated. Next, the support vector machine model is used to predict the class Ⅰ cell load data under coarse granularity. Finally, the class Ⅰ cell load density equalization coefficient is determined to calculate the classified load density index. With the local information the class Ⅱ cell load forecasting value is obtained, thereby meshing the results of spatial load forecasting. The engineering example verifies the practicability and effectiveness of this method.

关 键 词：空间负荷预测 地理信息系统 模糊信息粒化 支持向量机 网格化 

分 类 号：TM721[电气工程—电力系统及自动化]