Small sample-oriented power grid fault plan entity identification method and system
The invention discloses a small sample-oriented power grid fault plan entity identification method and system, and the method comprises the steps: obtaining historical fault data, designing a sentence pattern template, and achieving the first sample expansion; replacing a sentence pattern structure...
Gespeichert in:
Hauptverfasser: | , , , , , , , |
---|---|
Format: | Patent |
Sprache: | chi ; eng |
Schlagworte: | |
Online-Zugang: | Volltext bestellen |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
Zusammenfassung: | The invention discloses a small sample-oriented power grid fault plan entity identification method and system, and the method comprises the steps: obtaining historical fault data, designing a sentence pattern template, and achieving the first sample expansion; replacing a sentence pattern structure through a syntactic analysis technology to realize second sample expansion; unsupervised fine tuning training is carried out on the power text, and a power field Bert model is constructed; vectorizing the power grid fault plan text by using a power field Bert model; and identifying a key entity in the plan based on a deep recurrent neural network model. According to the invention, the information identification effect of the power grid fault handling plan is better.
本发明公开了一种面向小样本的电网故障预案实体识别方法及系统,包括:获取历史故障数据,设计句式模版,实现第一次样本扩充;通过句法分析技术,替换句式结构,实现第二次样本扩充;通过对电力文本进行无监督微调训练,构建电力领域Bert模型;利用电力领域Bert模型将电网故障预案文本向量化;基于深度循环神经网络模型识别预案中的关键实体。本发明对电网故障处置预案的信息识别效果更好。 |
---|