Abstract: A new intelligent question and answer platform for power system operation and maintenance knowledge is designed based on Baidu PaddlePaddle. Using natural language processing technology and FAISS as an indexing tool, SQL is used to store raw documents and metadata. The vector of text is constructed using FAISS indexing, and a practical intelligent Q&A system for power operation and maintenance is developed. The system is deployed using the Flask lightweight Web application framework to achieve online automatic Q&A, guiding users to carry out power operation and maintenance work based on the returned answers.