Intelligent Q&A System for Power Dispatching Based on Retrieval Augmented Generation

The knowledge of power dispatching is huge in scale and scattered in storage, making it time-consuming and laborious for operators to fully learn and utilize the knowledge documents. This paper proposes a intelligent Question and Answer (Q&A) system for power dispatching based on Retrieval Augmented Generation (RAG) technology. By optimizing the basic RAG for the characteristics of electric power documents and improving the text loading, chunking and retrieval methods, a set of intelligent Q&A system for electric power dispatching with accurate answers and reliable contents is constructed. And through experiments, it is verified that this system has a significant improvement in the electric power domain Q&A tasks compared with the basic RAG Q&A system.