Construction of a Multimodal Knowledge Graph for Power Grid Construction Safety Based on Large Language Models

In response to the difficulties of safety management and the complexity of information at power grid construction sites, a multimodal knowledge graph construction method based on large language models is proposed. Data from the construction site is collected and filtered, and an ontology for safety management at the construction site is constructed. The ontology is then used as retrieval augmented generation(RAG) for assistance, enabling multimodal large model image extraction, resulting in structured data in the power grid safety field. Finally, the extracted results are displayed using a graph database, completing the construction of the multimodal knowledge graph. The constructed knowledge graph includes multimodal data from the construction site, allowing for quick querying of on-site safety incidents, providing safety managers with a valuable tool for site management.