电网故障处置预案文本中的命名实体识别研究

电网故障处置预案是电网故障处置的重要参考,对电网故障处置预案文本中各类电力设备、名称编号等细粒度的关键实体信息进行抽取,是实现计算机学习理解预案内容并进一步支撑故障处置智能化的重要基础。文中提出一种基于深度学习的电网故障处置预案文本命名实体识别技术,首先采用字向量表征预案文本,然后将注意力机制以及双向长短期记忆网络相结合,有所侧重地提取实体词深层字符特征,最后采用条件随机场求解最优序列化的标注。算例表明:文中所提预案文本命名实体识别模型不依赖人工特征,能够自动高效地提取文本特征,准确识别预案文本中细粒度的实体词,满足预案文本中关键实体信息精确定位和识别的要求。

Named entity recognition in power fault disposal preplan text

Power grid fault disposal preplan is an important reference for power grid fault disposal. So extracting fine-grained key entity in-formation such as power equipments from the preplan is an important basis for the computer to understand the content and further support the intelligent disposal. This paper proposes a named entity recognition technology for power grid fault disposal preplan based on deep learning. Firstly, the word vector is used to represent the plan text. Then the word vector features are extracted by combining the attention mechanism and the bidirectional long and short-term memory network. Finally, the optimal serialization annotation is solved by the conditional random field. The example shows that the proposed entity recognition model can auto-matically and efficiently extract text features, thus accurately identifying entity words in the preplan. It proves that the model can meet the requirement of extracting key entity information in the preplan better than the other commonly used models.