Probabilistic Health Index-Based Apparent Age Estimation for Power Transformers

Abstract

This paper proposes a probabilistic health index-based method for estimating the apparent age of power transformer. Compared with the conventional weighted-score-sum based health index, the probabilistic health index is calculated as a data fusion result of kinds of transformer condition monitoring data through a constructed Bayesian belief network. The regression result of such a probabilistic health index is then applied to estimate the apparent age of the transformer through a few steps listed in the paper. The apparent age not only embodies an overall health status a transformer but also helpful for sorting a transformer fleet based on the estimated apparent age or even make it easy to make comparisons between transformer fleets. The estimated apparent age can be taken as a reference for power utilities to prioritize transformers and pay attention to the unit who owns the maximum apparent age among a fleet, thus helps to schedule replacement plans. Case studies with different transformers verify the usability and prove the advantages of the proposed method.

Keywords

• Apparent age
• Bayesian belief network
• condition monitoring data
• health index
• power transformer