Construction and Optimization of Knowledge Base for Commissioning Plans of New Power System Devices

To address the challenge of efficiently commissioning new equipment amid grid expansion, this study proposes an intelligent start-up plan generation method based on knowledge graphs. The approach first employs a bidirectional Long Short-Term Memory-Conditional Random Field model to extract entities such as equipment and operations from historical plans. This extraction process enables the construction of a structured knowledge base. Subsequently, it enhances plan standardization and reliability through multi-source data fusion and credibility analysis optimization logic. The results show that the named entity recognition model used in the study achieves an F1 score of 0.99, with an entity recognition accuracy of 92.5% for knowledge related to new power system devices commissioning plans. Testing of the proposed knowledge base reveals a knowledge coverage rate of 93.8%, which is 10.4% higher than that of traditional methods. The efficiency of plan generation significantly improves, with a rule compliance rate reaching 97%. The study provides a feasible pathway for transforming grid startup procedures from reliance on manual experience to data-driven intelligence.