Intelligent Troubleshooting System for SMT Manufacturing Using Web Scraping, GraphRAG and Knowledge Graphs

Mahr F, Franke J, Ockel M (2026)

Publication Type: Conference contribution

Publication year: 2026

Publisher: Institute of Electrical and Electronics Engineers Inc.

Pages Range: 361-362

Conference Proceedings Title: 2026 International Conference on Electronics Packaging and Hybrid Bonding Symposium, ICEP-HBS 2026

Event location: Hiroshima JP

ISBN: 9784991483516

DOI: 10.23919/ICEP-HBS69241.2026.11550728

Abstract

This paper presents the development of a GraphRAG-powered intelligent troubleshooting system for Surface Mount Technology (SMT) production environments. The system integrates multi-engine web scraping, Large Language Model (LLM) processing with Leiden algorithm-based community detection, and hierarchical query processing to create an effective knowledge retrieval architecture. Implemented at a Siemens Factory, the framework addresses critical challenges in technical availability by automatically extracting, processing, and organizing troubleshooting knowledge from distributed online sources. Evaluation results demonstrate a mean composite score of 0.74 with 83% of responses classified as high-quality. The system achieves 78.2% semantic similarity with expert responses. An adaptive keyword generation module enables continuous knowledge enhancement. This research establishes a reference architecture for AI-assisted manufacturing support systems.

Authors with CRIS profile

Felix Mahr Lehrstuhl für Fertigungsautomatisierung und Produktionssystematik (FAPS) Jörg Franke Lehrstuhl für Fertigungsautomatisierung und Produktionssystematik (FAPS) Manuela Ockel Lehrstuhl für Fertigungsautomatisierung und Produktionssystematik (FAPS)

How to cite

APA:

Mahr, F., Franke, J., & Ockel, M. (2026). Intelligent Troubleshooting System for SMT Manufacturing Using Web Scraping, GraphRAG and Knowledge Graphs. In 2026 International Conference on Electronics Packaging and Hybrid Bonding Symposium, ICEP-HBS 2026 (pp. 361-362). Hiroshima, JP: Institute of Electrical and Electronics Engineers Inc..

MLA:

Mahr, Felix, Jörg Franke, and Manuela Ockel. "Intelligent Troubleshooting System for SMT Manufacturing Using Web Scraping, GraphRAG and Knowledge Graphs." Proceedings of the 25th International Conference on Electronics Packaging, held in conjunction with the IEEE EPS Hybrid Bonding Symposium, ICEP-HBS2026, Hiroshima Institute of Electrical and Electronics Engineers Inc., 2026. 361-362.

BibTeX: Download