Application of Neural Network Analysis Based on Bayesian Regularization in Crosstalk of Cable

Abstract

In this paper, an artificial neural network based on Bayesian regularization training function is constructed for the internal twisted pair electromagnetic crosstalk system of aircraft. The crosstalk coupling current on the disturbed line is used as the training data of neural network and the coupling current is calculated by multi-conductor transmission line theory. The prediction results show that the prediction value of neural network based on Bayesian regularization training function is close to the input value, and the regression analysis shows that the prediction reliability of network is high, so it can be applied to the prediction of twisted pair electromagnetic crosstalk system.