摘要
为了能够更加及时有效地组织、管理和利用电力系统中的海量数据,需要利用知识图谱技术将其转化为电力相关知识,以帮助电力调度人员进行决策。本文在已有数据库的基础上,通过使用长短期记忆网络(long short-term memory,LSTM)+条件随机场(conditional random field,CRF)结构的框架进行电力知识的实体抽取,再利用脉冲耦合神经网络(pulse coupled neural network,PCNN)神经网络模型对数据库中的实体进行关系抽取,提取相关的电力知识,有效地改善了知识抽取的准确性与抽取速度。
In order to organize,manage and use enormous data of these power system data more timely and effectively,it is necessary to use knowledge map technology to transform them into power related knowledge to help power dispatchers make decisions.Based on the existing database,the framework of long short-term memory(LTSM)+conditional random field(CRF)structure is adopted to extract the entity of power knowledge,and then the pulse coupled neural network(PCNN)neural network model is applied used to extract the relationship between the entities in the database,and extract the relevant power knowledge,which effectively improves the accuracy and speed of knowledge extraction.
作者
李世明
LI Shiming(Power Dispatching Control Center,China Southern Power Grid Co.,Ltd.,Guangzhou 510000,China)
出处
《应用科技》
CAS
2023年第4期79-83,共5页
Applied Science and Technology
