Regression Model Extractions of a T-Equivalent Circuit Modelling for Medium-Length Transmission Line Based-on the Parametric Simulation Approach

Authors : Selami Balcı, Mustafa Akkaya

Pages : 1649-1658

Doi:10.2339/politeknik.1456959

View : 113 | Download : 165

Publication Date : 2024-09-25

Article Type : Research Paper

Abstract :In medium-length power transmission line models, the difference between the end-of-line and head-of-line voltage can be calculated with classical mathematical expressions. However, since the line parameters are not linear, these calculations can be approximated according to certain assumptions. The parametric data analysis approach proposed in this study obtained a data set for different variations by changing the line length and line parameters (transmission line specific parameters such as resistance, inductance, and capacitance) with certain steps. Then, using this data set, a classification is made with machine learning. In addition, data analysis is carried out with the end-of-line voltage value graphs obtained with different line parameters and the proposed approach is verified by constructing a test simulation circuit of a three-phase 200 km length with 154 kV line voltage value. Thus, a parametric simulation study has been presented, especially in electrical engineering education. In addition, Support Vector Regression (SVR) and Decision Tree Regression (DTR) models in the field of machine learning were used to measure the consistency of the data set created for 5 pF, 8 pF and 10 pF capacity values. With the figures and numerical data presented comparatively, it is clearly seen that the Long Short-Term Memory (LSTM) algorithm produces more successful scores in all three categories. In this context, the prediction accuracy was between 97% and 98% with DTR, while the accuracy was between 81% and 85% with SVR. Thus, prediction results in the range of 98% - 99% were obtained in the LSTM model.
Keywords : İletim hattı parametreleri, T eşdeğer devre modellemesi, DTR, SVR, LSTM

ORIGINAL ARTICLE URL