Akademik Veri Yönetim Sistemi
Journal of Energy Systems, cilt.7, sa.2, ss.199-211, 2023 (Scopus)
The issues related to integrating these systems into the grids continue to gain importance with the increasing use and importance of renewable energy sources. Therefore, the importance of power distribution transformers is increasing. Besides, these power distribution transformers are connected to the grid with power electronics circuits and inverters. Considering the modular inverter structures, ease of maintenance, and connection, three-level T-type inverters are chosen for this study. The secondary output voltage of the power transformer is estimated by using circuit parameters such as the dead time of the inverter circuit, PWM switching frequency, and modulation rate. Based on the finite element analysis analysis according to the selected parameters, 810 data are obtained with time-dependent parametric analysis. The adaptive neuro-fuzzy inference system model is constructed by considering the simulation data to estimate the secondary output of the power transformer of these parameters. In the training phase of the model, 648 randomly selected data from 810 data obtained by ANSYS-Electronics/Simplorer are used. The remaining 162 data are used in the testing process to measure system performance. As a result of the analysis made by ANFIS, the Root Mean Square Error (RMSE) error is found as 2.475%. Since the values obtained in the estimation process of the study are very close to the simulation values, the ANFIS method can be used as an estimation method that will give accurate results during the design phase.