SMART MANUFACTURING EXECUTION SYSTEM: REVOLUTIONIZING INDUSTRY WITH IOT AND AI (MODULE: CHATBOT)

Abstract

Manufacturing Execution Systems (MES) are essential for real‐time process monitoring and control in Industry 4.0 environments. Despite advanced IoT integration, maintenance technicians still depend on manual document searches through operator manuals, maintenance logs, and error‐code tables leading to prolonged troubleshooting, inconsistent results, and expensive downtime. To address these challenges, we present a fully on‐premise, Retrieval‐Augmented Generation (RAG) chatbot module that delivers contextually relevant procedural guidance directly within the MES. Our system preprocesses multimodal technical documents including PDFs, tables, and schematics using PyMuPDF, Camelot, Tesseract OCR, and BLIP image captioning, and partitions the content into semantically coherent, overlapping text chunks. A hybrid retrieval engine fuses BM25 keyword matching with dense‐vector similSarity search (via FAISS), and the top candidates are passed to a locally hosted Qwen3:1.7b language model through LangChain for answer generation. Importantly, all components run entirely on‐premise without any external API calls, preserving data privacy and eliminating subscription costs. Evaluation on real‐world machine manuals demonstrates sub‐second average response latency and over 90 % retrieval accuracy, significantly reducing reliance on printed manuals and accelerating maintenance tasks. Future work includes support for multilingual manuals and session‐memory for follow‐up queries.