With the rapid development of natural language processing technology, large language models have shown strong performance in various tasks. The retrieval-generation architecture is proposed to further improve the generation quality of the model while reducing the investment of computing resources. This paper proposes a RAG-based optimization method for large language models, including system complexity and efficiency optimization, knowledge base quality and maintenance, retrieval and generation results consistency, etc., which provides an effective solution for the optimization of large language models and has certain application value.