Automatic Generation Method of GIM Model Audit Rule Based on LLM

Substations widely employ Grid Information Model (GIM) for 3D design data delivery. The accuracy of GIM model is important for design, operation, and maintenance decision-making. Before transferring to various participators, audits should be conducted to ensure integrity and compliance of GIM model. However, conventional manual auditing methods have low efficiency and high costs. Moreover, the lack of computer-interpretable rules poses a significant challenge to the automated auditing. This study proposes an automatic generation method of GIM model audit rules. A standardized JSON structure encoding is proposed to describe the auditing rules. Based on the emergent capability and embedded vectors of Large Language Model (LLM), this study automates the conversion of auditing rules described in natural language into JSON rules. The effectiveness and applicability of this method are validated through case study. This method can enhance the efficiency of GIM model auditing, reduce errors caused by human intervention.