双碳背景下配电网智能终端非结构化信息抽取方法  

Unstructured Information Extraction Method of Intelligent Terminal in Distribution Network under the Background of Double Carbon

作　　者：李亚楠 LI Yanan(State Grid Shijiazhuang Electric Power Supply Company,Shijiazhuang 050019,China)

机构地区：[1]国网河北省电力有限公司石家庄供电公司,河北石家庄050019

出　　处：《微型电脑应用》2024年第5期183-186,共4页Microcomputer Applications

摘　　要：非结构化信息是配电网智能终端运行的基础与依据,但是其无序化的属性、较低的数据抽取效率影响了应用效果,为此,提出双碳背景下配电网智能终端非结构化信息抽取方法。应用面向对象建模技术与UML建模语言,构建配电网智能终端信息模型;基于Rocchio算法思想制定非结构化信息检索流程,在信息集合中分离出非结构化信息;采用正向最短编辑距离泛化处理非结构化信息,完成非结构化信息聚类;通过Bi-LSTM-CRF模型标注并抽取用户需求的非结构化信息,实现了配电网智能终端非结构化信息的抽取。实验数据表明,应用提出方法获得非结构化信息抽取时间最小达到3.6 s,抽取准确率数值高于0.80,召回率数值低于0.17,F 1数值低于0.28,充分证实了提出方法非结构化信息抽取效率与精度较高。Unstructured information is the basis for the operation of distribution network intelligent terminal,but its disordered attribute and low data extraction efficiency affect the application effect.Therefore,an unstructured information extraction method of distribution network intelligent terminal under the background of double carbon is proposed.This paper applies object-oriented modeling technology and UML modeling language to build the information model of intelligent terminal of distribution network,formulate the unstructured information retrieval process based on the idea of Rocchio algorithm,separate the unstructured information from the information set,use the forward shortest editing distance generalization to process the unstructured information,complete the unstructured information clustering,and mark and extract the unstructured information required by users through Bi-LSTM-CRF model.The unstructured information extraction of distribution network intelligent terminal is realized.The experimental data show that the minimum extraction time of unstructured information obtained by the proposed method is 3.6 s,the extraction accuracy is higher than 0.80,the recall is lower than 0.17,and the F 1 value is lower than 0.28,which fully confirms the high efficiency and accuracy of unstructured information extraction of the proposed method.

关 键 词：配电网 非结构化信息 智能终端 双碳背景 信息抽取 

分 类 号：TP31[自动化与计算机技术—计算机软件与理论]