Parameter estimation and validation of power transformers top oil temperature model by applying genetic algorithms

this paper presents a technique based on genetic algorithms for the parameter estimation and validation of the power transformers top oil temperature model proposed by lesieutre [1]. for such aim, data are used in on-line diagnosis and monitoring systems, installed in a 100 mva 230/115/24 kv oa/fa/foa transformer of barquisimeto substation at enelbar, venezuela since the year 2003. the objective of this work is to compare mistake reduction between the model and the top oil temperature measurement when their parameters estimation is considered by genetic algorithms and least-squares. the parameters estimation by genetic algorithms evidence better results of the model, which improves its performance as a power transformer diagnosis tool.