基于大数据挖掘的客户感知监控预警模型  

Customer perception monitoring and early warning model based on big data mining

作　　者：田诺 宫立华 朱龙珠 TIAN Nuo;GONG Lihua;ZHU Longzhu(State Grid Corporation of China Customer Service Center,Tianjin 300309,China)

机构地区：[1]国家电网有限公司客户服务中心,天津300309

出　　处：《电子设计工程》2024年第2期134-137,142,共5页Electronic Design Engineering

摘　　要：为了及时发现安全隐患,减少预警数据与真实数据存在的偏差,结合大数据挖掘方法构建客户感知监控预警模型。根据客户感知监控数据挖掘关联规则,捕捉电压偏离度、耗损量、电压不平衡率、电流不平衡率指标。结合动态聚类方法,更新客户感知监控数据集群。以聚类中心波动小于设置门限为基准判定收敛性,并计算不同层次间连接权值。在满足收敛性条件下构建监控预警模型,并进行动态预警和划分预警等级,实现客户感知监控预警。实验结果表明,该模型与标准值最大相差0.005 V、0.12 A,说明使用该模型预警范围精准,能够为客户提供精准预警数据。In order to find security risks in time and reduce the deviation between early warning data and real data,a customer perception monitoring early warning model is built by combining big data mining method.According to the customer perception,monitoring data mining association rules capture the indicators of voltage deviation,consumption,voltage unbalance rate and current unbalance rate.Combined with the dynamic clustering method,update the customer perception monitoring data cluster.The convergence is judged based on the cluster center fluctuation less than the set threshold,and the connection weights between different levels are calculated.Under the condition of meeting the convergence,the monitoring and early warning model is constructed,and the dynamic early warning and classification of early warning levels are carried out to realize the customer perceived monitoring and early warning.The experimental results show that the maximum difference between the model and the standard value is 0.005 V and 0.12 A,indicating that the early warning range of the model is accurate and can provide accurate early warning data for customers.

关 键 词：大数据挖掘 客户感知 监控预警 动态聚类 

分 类 号：TN01[电子电信—物理电子学] TP399[自动化与计算机技术—计算机应用技术]