Banca de QUALIFICAÇÃO: PEDRO HENRIQUE CASSIMIRO CALDEIRA

Uma banca de QUALIFICAÇÃO de MESTRADO foi cadastrada pelo programa.
DISCENTE : PEDRO HENRIQUE CASSIMIRO CALDEIRA
DATA : 15/07/2026
HORA: 13:30
LOCAL: Sala virtual: meet.google.com/pfh-zapq-foh
TÍTULO:

DGA-Based Diagnosis and Risk Assessment of Power Transformers with Integration of Operational Variables and a Data-Driven Approach

PALAVRAS-CHAVES:

Power Transformers, Dissolved Gas Analysis (DGA), Fuzzy Logic, Criticality Matrix, Asset Management.

PÁGINAS: 63
GRANDE ÁREA: Engenharias
ÁREA: Engenharia Elétrica
RESUMO:

The monitoring and diagnosis of incipient faults in oil-immersed power transformers are essential to ensure the reliability, operational safety, and longevity of electrical systems. In this context, Dissolved Gas Analysis (DGA) stands out as one of the main techniques used in predictive maintenance programs. However, traditional standard-based methods focus predominantly on identifying the type of fault, without incorporating, in a structured way, the risk assessment associated with the real operating conditions of the assets, which may lead to inconsistent or inconclusive decisions.

Therefore, this work aims to evaluate the reliability and consistency of fault diagnosis in power transformers through the development of an integrated methodology based on Fuzzy Inference Systems and risk analysis. The proposed methodology utilizes a real dataset of chromatographic analyses collected between 2023 and 2025 from an asset park of a multinational company in the Brazilian agro-industrial sector.

The structured framework relies on a two-level fuzzy inference process: the first level determines the severity of the operational condition by integrating the asset’s age, oil temperature, moisture content, and the laboratory’s chromatographic diagnosis; the second level operates on a 5 × 5 criticality matrix aligned with ISO 31000 principles, combining the calculated severity with a frequentist probability based on Rogers’ Ratio Method.

This dual fuzzy process mitigates the discontinuities of discrete classifications, generating a continuous indicator called Risk Degree (RD), which classifies assets into four intervention priority levels. Expected results demonstrate that incorporating operational variables, especially age and moisture, significantly changes the risk classification of assets with similar gas signatures, proving that isolated DGA analysis is insufficient to represent the actual criticality of transformers. As a main contribution, the model provides a quantitative and robust decision-making support tool for asset management, bringing technical diagnosis closer to the practical needs of maintenance engineering

MEMBROS DA BANCA:
Presidente - 2362923 - GIOVANI BERNARDES VITOR
Interno - 1521733 - CLODUALDO VENICIO DE SOUSA
Interno - ***.230.456-** - LUIZ CARLOS BAMBIRRA TORRES - UFOP
Externo ao Programa - 2225499 - IVAN PAULO DE FARIA - UNIFEI