An advanced retrieval-augmented generation system for manufacturing quality control

Abstract

The rise of Large Language Models (LLMs) with generative artificial intelligence has revolutionized the development of knowledge-based systems, enabling intuitive interactions through natural language. This paper explores the implementation of an advanced Retrieval-Augmented Generation (RAG) system, designed to improve manufacturing quality control by utilizing the capabilities of LLMs, particularly OpenAI’s GPT models. We focus on the ceramic tile manufacturing process, where the system retrieves and analyzes specialized bibliographic sources to diagnose defects and propose solutions. In addition to core RAG functionalities, the system incorporates tailored pre-processing and post-processing mechanisms to optimize document retrieval and response generation. The system’s effectiveness in solving quality issues is demonstrated through its application in identifying defect causes and generating actionable solutions, significantly improving non-conformities management. This approach not only streamlines troubleshooting but also enhances the quality control system, providing a comprehensive, scalable tool for manufacturers.