基于DeepSeek⁃R1的纺纱工艺专家系统框架模型及其应用OA

In response to the problems that plaguing the high-quality development of spinning enterprises in Xinjiang,such as strong dependence on empirical knowledge,insufficient process optimization and shortage of professional talent,the establishment of a localized spinning process expert system based on DeepSeek-R1 large language model and retrieval-augmented generation(RAG)technology was explored,aiming to promote the intelligent production transformation and upgrading of enterprises.DeepSeek-R1 model was deployed locally in a lightweight manner by using Ollama.Vector knowledge base was constructed by RAG technology for realizing knowledge retrieval.RAG service was packaged by Docker containerization technology and communicated with Ollama through HTTP interfaces,then a modular architecture was formed.Yarn strength CV indicator monitoring was selected as example,the preliminary application practice indicated that,the root causes for lower process capability in compact spinning production can be rapidly identified and analyzed by the system,and guidance with traceable causal chains could be generated,the professional knowledge recall capability was better.The expert system can transform fragmented process knowledge into reusable and interpretable structured decision support tools,enhancing the efficiency and transparency of process diagnosis,and providing technology support for intelligent diagnosis and process decision-making in enterprises spinning production.