Enhanced Analysis of Electric Field Distribution in High Voltage Insulators Using Advanced ANN

The design of insulators fundamentally relies on the accurate assessment of the electric field distribution on their surfaces. Traditionally, this assessment is conducted using the Finite Element Method (FEM), a technique that involves segmenting the insulator surface into distinct, non-overlapping regions. This process is notably time-consuming and demands extensive numerical computations. However, integrating Artificial Neural Networks (ANN s) offers a simplified approach to modeling these complex relationships. When an ANN is adequately trained, it significantly reduces the time and computational intensity required to determine the electric field distribution on insulators. This paper introduces an ANN model that utilizes the back-propagation algorithm to predict the electric field distribution on various insulators under high-voltage (HV) conditions. The model iteratively adjusts its structure to achieve an optimal configuration that minimizes the mean square error. The inputs for the ANN are the dimensional coordinates of selected nodes on the insulator surface and the insulator's voltage level, with the electric field distribution as the output, derived from FEM data. The study uses 600 data patterns, with 90% allocated for training the ANN and the remaining 10% for testing. The results acquired from the ANN model are juxtaposed with those obtained through the FEM approach, demonstrating the ANN's effectiveness and validating its performance in predicting electric field distributions on insulators.