A novel approach based on Simulated annealing coupled to artificial neural network for 3-D electric-field optimization

This paper deals with the optimization of stress distribution on and around electrode-spacer arrangements for ensuring economical and higher reliability of gas-insulated systems. The optimization technique adopted in this paper is the artificial neural network-aided simulated-annealing (SA) algorithm. By coupling a trained neural net with the annealing algorithm, the execution speed of the latter is greatly enhanced to evaluate the optimum values for the design parameters of the electrode-spacer arrangements compared to calculating the cost function via the entire process for field calculation at every move of the optimization algorithm. The convergence of the optimization algorithm has also been compared statistically with the genetic algorithm. The results presented in this paper show that optimized electrode-spacer contours have been obtained with an acceptable degree of accuracy using the neural-network-aided SA algorithm. The statistical analysis shows a promising result for the proposed method.