Establishing a Model for Predicting Material Wear and Crack Initiation Based on BP Neural Network: Taking Ceramic Materials as an Example

In order to establish a model for predicting partial wear and crack initiation of disc suspension porcelain insulators, this article presents a novel approach for predicting porcelain wear and crack initiation by utilizing a BP neural network and non-equidistant grey model. The proposed method is based on an analysis of actual operating conditions and measurements of wear and crack initiation in porcelain components. The goal is to improve the accuracy of wear and crack prediction in porcelain components under different operating conditions. The prediction model utilizes raw experimental data measured at unequal time intervals to predict ceramic wear and crack initiation. The measured ceramic wear and crack initiation data are used as model training samples to obtain ceramic prediction results. After model training, it can be seen that after optimizing the residual sequence using the BP network, the prediction results of porcelain wear using the GM-BP model are basically consistent with the actual prediction curve, with a relative error of 0.091%. Compared with the prediction results of the GM model, the accuracy has improved by 31.6%. The GM-BP model also has good prediction accuracy in predicting crack initiation in porcelain components, which has important reference significance for ensuring the safe operation of the power system.