Gas insulated switchgear (GIS) is an important equipment in substations. With the promotion of digital power grids and the transformation and upgrading of substations, the problems of low efficiency and high cost in traditional operation and maintenance methods have put forward higher requirements for operation and maintenance personnel. Based on this, the paper proposes a construction method of GIS operation and maintenance knowledge graph based on ELECTRA algorithm, aiming to achieve more intelligent and efficient maintenance of GIS equipment. Firstly, this article constructs a training dataset using publicly available data and sample generation methods, and, on this basis, constructs an ontology layer of the knowledge graph. Then, the ELECTRA-BiLSTM-CRF model and ROBERTA model were used to extract entities and relationships from the dataset, and the entity layer of the knowledge graph was constructed. Finally, the construction of the GIS equipment operation and maintenance knowledge graph was achieved. Through comparative experimental analysis, the F1 values of the proposed method in entity and relationship extraction are 90.01% and 86.04%, respectively, 2.73%–6.12% higher than other methods, which has better progressiveness and popularization significance.