Fault Diagnosis in Power Transformers Using Acoustic and Electromagnetic Sensing Techniques

• Grant number: 24/17128-4
• Support Opportunities: Regular Research Grants
• Start date: March 01, 2025
• End date: February 29, 2028
• Field of knowledge: Engineering - Electrical Engineering
• Principal Investigator: Bruno Albuquerque de Castro
• Grantee: Bruno Albuquerque de Castro
• Host Institution: Faculdade de Engenharia (FE). Universidade Estadual Paulista (UNESP). Campus de Bauru. Bauru , SP, Brazil
• Associated researchers: Danilo Hernane Spatti ; Jorge Alfredo Ardila-Rey ; Stefan Tenbohlen

Abstract

The interest in predictive failure diagnosis systems for power transformers has grown significantly in recent years. Monitoring allows for a high degree of control over the condition of these electrical machines, contributing to substantial savings in maintenance and helping to prevent outages. In this way, these methodologies support sustainable development and improve the quality of a country's power supply indices, and consequently, the well-being of people. High-voltage equipment insulation systems, such as power transformers, are subject to multiple critical factors in their operations. These factors, which may be electrical, thermal, mechanical, or environmental, lead to a gradual degradation of the machine's physical-chemical insulation properties, which may result in total failure. Prior to the outage, partial discharges (PD) commonly occur in the insulation components. When PD occur, it produces heat, light, electromagnetic, and acoustic waves, accelerating the degradation of the machine's insulation system. Among the various non-destructive techniques for power transformer monitoring, two of the most promising are acoustic emission, which uses piezoelectric transducers to detect acoustic waves originating from faults, and UHF sensing, which uses antennas to detect electromagnetic waves emitted by faults. However, the use of piezoelectric transducers and antennas for diagnosing fault types in transformers remains an open topic in the literature. Although monitoring techniques have been studied for several decades, practical issues for real-world applications still need to be investigated. The current challenge is to develop sensors and new feature extraction techniques that can identify the type and the severity of the failure and its localization. Additionally, there is a need to study the feasibility of these techniques in real environments, which may be subject to environmental noise that could limit the applicability of these systems. Therefore, the goal is to develop alternative methods capable of incorporating both fault type classification, as each fault requires a specific maintenance action, and severity assessment, especially in real-world applications where environmental noise is present and can interfere with measurements. It is also a research project involving Unesp, Brazil, the University of Stuttgart, Germany, Federico Santa Maria Technical University, Chile, and the University of São Paulo, Brazil. (AU)