Condition Evaluation of Dry-type Transformer Based on High-dimensional Random Matrix Theory

Abstract

Epoxy dry-type transformer plays a key role in the offshore oil platform power system. The normal operation of dry-type transformers faces many challenges, mainly due to the long maintenance period, high reliability requirements and complex offshore power requirements. At the same time, the offshore power system has formed a big data environment. In this context of power system, big data analysis methods are urgently needed. Based on the high-dimensional random matrix theory, this paper firstly considers various factors which have influence on the state of dry-type transformers to construct a condition evaluation matrix, and then analyzes the eigenvalue distribution of the condition evaluation matrix which was formed by time series data. In order to reflect changes in eigenvalue distribution, the mean spectral radius (MSR) was introduced. Through it, we can find the trend of key performance changes, and detect abnormalities in key performance of equipment in time. Finally, the effectiveness of the proposed method is illustrated by an example.