Análisis predictivo de fallas internas en transformadores de potencia utilizando indicadores de condición provenientes de un sistema de monitoreo en línea

Abstract

This thesis focuses on the predictive analysis of internal failures in power transformers, based on the study of condition indicators obtained through online monitoring systems installed in electrical substations. Variable operating conditions, such as thermal fluctuations, the presence of moisture in the oil, and dielectric stresses, increase the probability of progressive deterioration of the internal insulation, making it necessary to strengthen predictive maintenance strategies to guarantee the reliability and continuity of the electrical supply. The research is based on the continuous collection of relevant operating parameters, such as oil temperature, moisture content, dissolved gases, partial discharges, and insulation aging rates. These indicators allow for the evaluation of the transformer's actual behavior during normal operation, overcoming the limitations of periodic inspections and isolated measurements, which often do not reflect the dynamic state of the equipment. Based on the information recorded by the monitoring system, the various condition indices are classified and correlated with previously occurring thermal, mechanical, and dielectric events. This comprehensive approach allows for the identification of incipient degradation patterns and the establishment of relationships between variables that, analyzed individually, might go unnoticed by operating personnel. The expected results of this study contribute to improving maintenance management in power transformers, facilitating the early detection of internal faults, reducing operational risks, and optimizing economic resources associated with major repairs or unplanned replacements. Furthermore, this work contributes applied knowledge in the field of online monitoring and electrical asset diagnostics, strengthening predictive maintenance practices in electrical substations.