Comparison of some neural network algorithms used in prediction of XLPE HV insulation properties under thermal aging

Some Artificial neural network algorithms have been used to predict properties of high voltage electrical insulation under thermal aging in term to reduce the aging experiment time. In this paper we present a short comparison of the obtained results in the case of Cross-linked Polyethylene (XLPE). The theoretical and the experimental results are concordant. As a neural network application, we propose a new method based on Radial Basis Function Gaussian network (RBFG) trained by two algorithms: Random Optimization Method (ROM) and Back-propagation (BP).