Application of the Dielectric Permittivity Parameter of Crude Oil during Its Processing

Abstract

This study aims to address the issue of predicting the ice load on overhead power lines using neural network modeling. This issue was implemented on the basis of a forecasting model with a 12-8-1 architecture using the method of sliding time windows in the STATISTICA environment. The average relative error of the air temperature forecast was 12.9 %, which confirms the reliability of the forecast model and the feasibility of using neural network modeling to solve forecasting problems in the energy sector. The presented scientific research can be successfully used by operators of the power distribution system to adjust the operation of power grids in accordance with the meteorological situation along the overhead line route.