Abstract With the advancement of smart grid construction, substation equipment management faces challenges such as multisource heterogeneous data and complex fault correlation analysis. This paper proposes a decision-support method for substation intelligent butlers that integrates knowledge graphs and graph convolutional neural networks (GCN). By constructing a whole-station knowledge graph covering key equipment such as transformers and circuit breakers, it realizes structured representation and correlation analysis of equipment status. The Sentence-BERT model is introduced to address semantic redundancy in knowledge fusion, and transformer fault identification is achieved based on GCN. Experimental results show that this method achieves an overall accuracy of 92.8% in transformer fault diagnosis, significantly outperforming traditional deep learning models in both accuracy and interpretability, providing an efficient technical path for intelligent decision support in substations.
Guo et al. (2026) studied this question.