An Approach to Remaining Life Estimation for Power Transformers with Biased Data

Abstract Power transformers are among the most important equipment in power systems. Estimating the remaining life of transformers is an important issue for energy companies because of the need for planning maintenance and capital expenditures. Traditional methods fail to provide accurate results mainly because data is incomplete, and there is a significant lack of records of equipment scrapped before 2015. Given this, the study provides a new approach to life estimation for power transformers based on survival analysis. First, the data was selected from only the transformers that were still operating after 2015. Then, a Bayesian survival model with a truncated Weibull distribution is developed to fit the biased data. Parameters are, at last, estimated through maximum likelihood estimation. The experimental results indicate that the proposed method performs well.