ScholarWorks@동국대학교

초록

Power transformers are critical assets in modern power systems because they enable the efficient transmission and distribution of electricity to industrial, commercial, and residential users. Although these devices are engineered for durability, they are still susceptible to electrical, thermal, and mechanical stresses that may result in failures, potentially triggering costly outages and compromising grid reliability. Consequently, timely detection of internal and external faults is crucial for maintaining uninterrupted operation and prolonging the service life of power transformers. In recent years, Intelligent Prognostics and Health Management (iPHM) has become a transformative standard that enables a shift from traditional time-based maintenance to autonomous, condition-based maintenance strategies. This review explores the implementation of iPHM for power transformers, emphasizing the integration of advanced sensor systems with intelligent computational techniques, including machine learning (ML) and deep learning (DL). The review begins by presenting a summary of the principal sensing technologies, then addresses data preprocessing techniques and computational strategies for fault identification, severity evaluation, and the estimation of Remaining Useful Life (RUL). In addition, advanced methods like transfer learning and hybrid approaches are analyzed for their capability to enhance adaptability in varying operational conditions and to mitigate issues related to limited data. By consolidating developments in both sensing and prognostic technologies, this review further outlines ongoing challenges and proposes future research directions for improving the reliability and safety and operational lifespan of power transformers within contemporary energy infrastructures.

키워드