Abstract: There are a lot of complex data in the power dispatching operation system. Building the knowledge graph of power grid dispatching field is an important means to solve the lack of data relevance and low added value, and the named entity recognition of power grid dispatching field is one of the basic tasks of building the knowledge graph of power grid dispatching field. Aiming at the problems of poor applicability and slow model training speed of general domain named entity recognition method in the field of power grid regulation, this paper proposes a named entity recognition method in the field of power grid dispatching based on ERNIE-IDCNN-CRF model. This method establishes a unified model of words, phrases, entities and other information, introduces multi-source heterogeneous data knowledge to generate semantic vectors, and significantly enhances the semantic expression ability of the model. The experimental results show that the training speed of this method is significantly improved for the power dispatching corpus, and the F1 value of the entities in the power grid dispatching field reaches 85.09%, which can effectively identify the entities in the power grid dispatching field.