Research on Condition Assessment and Fault Diagnosis of Power Transformers

Abstract

Performing condition assessment and fault diagnosis for power transformers is an indispensable part to guarantee the safe and economical operations of the power grid. Condition monitoring is the basis of condition assessment and fault diagnosis, and the accuracy of collected data will ensure the accuracy of assessment and diagnosis from the source. During the construction of smart grids, the historical operating data of power transformers present various characteristics such as large quantity and numerous types, so that the condition assessment and fault diagnosis algorithms gradually transition from the threshold judgment method to the machine learning algorithm. In this paper, the methods of transformer condition monitoring in recent years are summarized, and some common research methods of power transformer condition assessment and fault diagnosis are outlined. Meanwhile, the applications of traditional algorithms and artificial intelligence algorithms are introduced. In addition, the existing challenges faced in this field are briefly analyzed, and an outlook on the main research directions for the future is accordingly provided.