A New Method for Predicting Crosstalk of Random Cable Bundle Based on BAS-BP Neural Network Algorithm

Abstract

Accurate analytical solution for the crosstalk of random cable bundle is difficult to obtain, but the limit of the crosstalk can be predicted. This paper proposes a method to predict the crosstalk of random cable bundle. Based on the idea of cascade method, the model takes into account the random rotation of the cross-section and the random transposition of the core. A neural network algorithm based on back propagation optimized by the beetle antennae search method (BAS-BPNN) is introduced to mathematically describe the random rotation of the cross-section. The elementary row-to-column transformation of the unit length RLCG parameter matrix is used to deal with the random transposition of the core. The discontinuity between segments generated by transposition is solved by introducing transition probability parameters. Finally, combined with the finite-difference time-domain (FDTD) algorithm, the crosstalk of the random cable bundle is obtained. The numerical experimental results show that the new method can reduce a lot of experimental work in the crosstalk problem of random cable bundle, and has higher accuracy and a wider frequency range.

Keywords

• Crosstalk
• random cable bundle
• beetle antennae search (BAS) algorithm
• back propagation neural network (BPNN) algorithm
• finite-difference time-domain (FDTD)
• multi-conductor transmission lines (MTLs)