Banca de QUALIFICAÇÃO: MATHEUS GOMES DINIZ ANDRADE

Uma banca de QUALIFICAÇÃO de MESTRADO foi cadastrada pelo programa.
STUDENT : MATHEUS GOMES DINIZ ANDRADE
DATE: 23/07/2025
TIME: 09:00
LOCAL: Remoto
TITLE:

A Multi-Agent Methodology for Knowledge Extraction in PROFIBUS Technical Standards

KEY WORDS:

Retrieval-Augmented Generation, Large Language Models, Model Context Protocol, Multiagent, Information Retrieval, PROFIBUS.

PAGES: 60
BIG AREA: Engenharias
AREA: Engenharia Elétrica
SUMMARY:

Industry 4.0 has transformed the manufacturing sector through digitalization, automation, and data sharing.
To support this integration, various communication protocols have been adopted, with fieldbuses standing out. However, while this variety offers flexibility, it also hinders interoperability, motivating research on protocol conversion and integration. PROFIBUS is one of the most widely used protocols in industrial automation, operating across multiple layers of the OSI model and supporting diverse production scenarios. Nevertheless, its maintenance poses challenges, especially in legacy systems where faults are difficult to diagnose. In this context, AI-based intelligent solutions have emerged by providing fast and contextualized access to technical documentation. The combination of Retrieval-Augmented Generation (RAG) and Large Language Models (LLMs) has significantly advanced question-answering systems, particularly in technical domains that require high precision and continuous adaptability, enhancing answer relevance without the need for retraining. However, their limitations have led to the exploration of agentic systems, which introduce autonomous agents capable of further improving the flexibility and performance of RAG architectures in complex and evolving scenarios. Given this context, this work proposes the development and evaluation of architectures that integrate multi-agent systems to optimize the retrieval and interpretation of information in technical documentation for the PROFIBUS protocol.

 

COMMITTEE MEMBERS:
Presidente - 2885532 - IVANOVITCH MEDEIROS DANTAS DA SILVA
Interno - 1153006 - LUIZ AFFONSO HENDERSON GUEDES DE OLIVEIRA
Interna - ***.640.764-** - MARIANNE BATISTA DINIZ DA SILVA - UFRN
Externo à Instituição - DENNIS BRANDAO - UniBs