A Case Study on the Use of a Specialized Chatbot in the Korean Power System

The increasing complexity of power systems has made traditional analysis methods less effective, leading to the application of artificial intelligence (AI) in the power system. This study develops a domain-specific chatbot using generative AI to interpret regulations related to distributed energy resources (DERs) in South Korea, trained on KEPCO’s &quotTechnical Standards for the Connection of Distributed Energy Resources to Distribution Systems.&quot The chatbot uses keyword-based retrieval and sentence generation to answer queries on terms like &quotsurge&quot &quotflicker&quot and &quotharmonics&quot. Challenges such as sentence coherence in complex queries were identified. To address this, the Llama-based model and keyword-based information retrieval methods are utilized. Despite these challenges, the chatbot shows strong potential for improving regulatory information access and user interactions in the power sector.