摘要
电网云平台在智能电网中占据着至关重要的位置,其需要接收、处理与分析海量的多属性数据,对其分类挖掘性能提出了极高的要求,故提出电网云平台多属性数据分类挖掘方法研究。基于深度自编码器提取电网云平台多属性数据特征,并通过信息增益对提取数据特征进行选择,从而降低数据特征维度。引入支持向量机构建多属性数据分类挖掘模型,制定独立支持向量机训练流程。将提取的多属性数据特征输入至训练好的模型中,即可获得最终的多属性数据分类挖掘结果。测试数据显示:提出方法应用后AUC值维持在较高水平(0.84~0.95),Kappa系数的最大值达到了1,充分证实提出方法具备较好的应用性能。
Grid cloud platform occupies a vital position in smart grid.It needs to receive,process and analyze massive multi-attribute data,which puts high demands on its classification and mining performance.Therefore,the research on classification and mining method of multi-attribute data on grid cloud platform is proposed.Multi-attribute data features of power grid cloud platform are extracted based on depth self-encoder,and the extracted data features are selected through information gain,thus reducing the data feature dimension.Support vector machine is introduced to build a multi-attribute data classification mining model,and the training process of independent support vector machine is formulated.Input the extracted multi-attribute data features into the trained model,and the final multi-attribute data classification mining results can be obtained.The test data show that the AUC value of the proposed method remains at a high level(0.84~0.95)and the maximum Kappa coefficient reaches 1,which fully proves that the proposed method has good application performance.
作者
胡雨
刘膨源
王玉玲
潘士通
杨君
HU Yu;LIU Pengyuan;WANG Yuling;PAN Shitong;YANG Jun(Beijing SGITG Accenture Information Technology Center Co.,Ltd.,Beijing 100052,China)
出处
《自动化与仪器仪表》
2025年第10期35-38,共4页
Automation & Instrumentation
基金
国家电网有限公司安徽省电力有限公司项目(B36817230151)。
关键词
多属性数据
支持向量机
分类挖掘
电网云平台
核函数选择
数据特征提取
multi-attribute data
support vector machine
classification mining
power grid cloud platform
kernel function selection
data feature extraction
