Özet:

As the demand for electricity increases, the importance of transformers, which are one of the most important equipment used in the transmission and distribution of this energy, also increases. One of the most important parameters that transformers need to be monitored in order to be able to work efficiently for a long term is the heat-spot temperature (HST) value. A variety of thermal models are used to determine this parameter. However, as the thermal events within the transformator are complex, thermal models that give the HST value are not developed. The aim of this study is to create a regression model to find the optimal parameter values used in the forecasting of maximum HST using three-level Box-Behnken design and genetic algorithm (GA) methods combined with the Response Surface Method-RSM. Thus, the HST value can be predicted to increase the life and productivity of the transformers. In the study, the 5kVA 380/220V single-phase dry-type transformator has prepared the experimental arrangement. The experiments were conducted taking into account three different current values (4, 8 and 12 A), current application time (3, 6 and 9 hours) and box-behnken design at medium temperature (20, 25 and 30 oC). The results of the experiment are Design Expert 11.0. A mathematical predictive model for HST was developed and the maximum HST was predicted by achieving the optimal values of the parameters. Later, GA was used for HST forecast using the mathematical prediction model obtained from RSM. At the end of the study, the estimated values obtained through both methods were compared and found that they were very close to each other.

Anahtar Kelimeler: