低电压预警监测智能决策系统设计与实现  

作　　者：李强 杜丰夷 范李平[1] 沈映彤 熊瑛 

机构地区：[1]国网宜昌供电公司,湖北宜昌443000 [2]三峡大学电气与新能源学院,湖北宜昌443000 [3]国网武汉供电公司,武汉430000

出　　处：《科技创新与应用》2024年第36期43-46,共4页Technology Innovation and Application

摘　　要：为解决目前多系统数据难融合、低电压研判不准确、低电压问题难根治和治理效率低等问题,该文搭建基于大数据挖掘技术的低电压智能分析预警诊断模型。构建基于多任务辅助学习的全景感知低电压监测模型,实现高效、准确的电力客户分群与低电压成因分析;提出基于卷积神经网络的命名实体识别模型,形成“实体-关系-实体”三元组的低电压知识图谱;应用低电压知识图谱,实现辅助决策自动生成,并通过平台进行可视化展示。结果表明,低电压预警监测智能决策系统涵盖供电指挥、配网生产、营销服务和发展规划等专业的业务全流程、全环节,可大幅提升供电服务水平和业务支撑能力。In order to solve the current problems of difficulty in integrating multiple system data,inaccurate low voltage judgment,difficulty in curing low voltage problems,and low governance efficiency,this paper builds a low voltage intelligent analysis,early warning and diagnosis model based on big data mining technology.A panoramic sensing low voltage monitoring model based on multi-task assisted learning is built to achieve efficient and accurate power customer grouping and low voltage cause analysis;a named entity recognition model based on convolutional neural network is proposed to form an"entity-relationship-entity"low voltage knowledge graph of the triple group;the low voltage knowledge graph is applied to realize automatic generation of auxiliary decisions and visually displayed through the platform.The results show that the low-voltage early warning and monitoring intelligent decision-making system covers the entire business process and all links of power supply command,distribution network production,marketing services,development planning and other majors,greatly improving the power supply service level and business support capabilities.

关 键 词：低电压 大数据 多任务辅助学习 卷积神经网络 知识图谱 

分 类 号：TM714[电气工程—电力系统及自动化]