An improved computational approach for thermal modeling of power transformers

Abstract

Summary Proposed IEEE/ANSI differential equation governing top oil temperature rise, and its improvements are widely used in top oil temperature prediction and parameter estimation for power transformer loss of life calculation and diagnostic systems. Approximately, all transformer thermal models employ backward Euler method to achieve related set of equations (or regressors) and use least square error estimator for parameter estimation. In this paper, we applied a new numerical integration methodology to solve the differential equation of thermal model. Related matrix format is calculated based on the numerical integration rectangle rule. Some comparison metrics are introduced to verify and compare the proposed computation methodology with conventional discretized backward Euler equation. The proposed method was tested on simulated top oil temperature using measured ambient temperature and load data in the field. Results showed a significant and meaningful improvement in top oil temperature prediction. Variations of calculated parameters, in the presence of noise, presented the superiority and advantages of the proposed computational methodology. Copyright © 2014 John Wiley & Sons, Ltd.