Application of an improved BP network for the prediction of insulation operating state based on GNBR algorithm

Abstract

The BP network is widely accepted as a technology offering an alternative way to capture nonlinear patterns to complex real-world problem. In this article, BP neural network is used to identify the nonlinear relationship between salt & ash density and insulator flashover voltage which is very important in insulation operate state monitoring. The salt density and ash density are main contamination for insulator surface. In different regions and different environment, salt density and ash density have different impact to insulator flashover voltage. In order to improve the BP network character, the GNBR algorithm is used in the training process. The improved BP network has better character such as optimum ability, fast speed and high recognition accuracy. Experimental results show that the designed improved BP network has good identification ability to predict the relationship between the insulator contamination and flashover voltage.