A New Method for Crosstalk Prediction Between Triple-twisted Strand (Uniform and Non-uniform) and Signal Wire based on CDBAS-BPNN Algorithm

Abstract

This paper proposes a novel crosstalk prediction method between the triple-twisted strand (uniform and non-uniform) and the signal wire, that is, using back-propagation neural network optimized by the beetle antennae search algorithm based on chaotic disturbance mechanism (CDBAS-BPNN) to extract the per unit length (p.u.l) parameter matrix, and combined with the chain parameter method to obtain crosstalk. Firstly, the geometric model and cross-sectional model between the uniform triple-twisted strand and the signal wire are established, and the corresponding model between the non-uniform triple-twisted strand and the signal wire is obtained by the Monte Carlo (MC) method. Then, the beetle antennae search algorithm based on chaotic disturbance mechanism (CDBAS) and backpropagation neural network (BPNN) are combined to construct a new extraction network of the p.u.l parameter matrix, and the chain parameter method is combined to predict crosstalk. Finally, in the verification and analysis part of the numerical experiments, comparing the crosstalk results of CDBAS-BPNN, BAS-BPNN and Transmission Line Matrix (TLM) algorithms, it is verified that the proposed method has better accuracy for the prediction of the model.