Decision Support System for Diagnosis of Power Transformers

Given the complexity of the Brazilian Electric Power System, it is important to ensure that equipments work in satisfactory conditions and with high levels of reliability. Within those equipments, power transformers are one of the most important assets, because they are expensive and functionally complex. Dissolved Gases Analysis (DGA) in mineral oil insulation is widely considered as an appropriate tool for the identification of faults in power transformers. This work proposes a system called SADTRAFOS to support maintenance managers' decisions regarding those faults. The system is composed of a fuzzy inference module for fault diagnosis, and of a decision support module that gives recommendations to maintenance managers. The experiments carried out with a real database attained more than 80% accuracy in the fault diagnosis and correct recommendations for the maintenance team.