LA-RAG:Enhancing LLM-based ASR Accuracy with Retrieval-Augmented Generation