Analysis of Crosstalk Problem in Multi-Twisted Bundle of Multi-Twisted Wire Based on BSAS-BP Neural Network Algorithm and Multilayer Transposition Method

Abstract

Twisted wire used in complex systems has the ability to reduce electromagnetic interference, but crosstalk within the wire is not easy to obtain. This paper proposes a method to predict the crosstalk of multi-twisted bundle of multi-twisted wire (MTB-MTW). A neural network algorithm based on back propagation optimized by the beetle swarm antennae search method (BSAS-BPNN) is introduced to mathematically describe the relationship between the twist angle of the wire harness and the per-unit-length (p.u.l) parameter matrix. Considering the symmetry of the model, the relationship between the unresolved angle of the BSAS-BPNN algorithm and the p.u.l parameter matrix is processed by using the multilayer transposition method. Based on the idea of the cascade method and the finite-difference time-domain (FDTD) algorithm in Implicit-Wendroff format, the crosstalk of the wire is obtained. Numerical experiments and simulation results show that the new method proposed in this paper has better accuracy for the prediction of the model. The new method can be generalized to the MTB-MTW model with any number of wires. All theories provide preliminary theoretical basis for electromagnetic compatibility (EMC) design of high-band circuits.