A New Method for Predicting Crosstalk of Hand-Assembled Cable Bundles

Abstract

Hand-assembled cable bundles are random harness whose crosstalk is difficult to obtain accurately. A crosstalk prediction method of hand-assembled cable bundles is proposed in this paper. The harness is modeled by means of the mean pseudo-random number based on the cascade method. The factors considered in the model include the random exchange of wires position in the wiring harness cross section and the random rotation of the cross section to the ground. A mathematical description of the random exchange of wires position is made by using the row and column transformation of the per unit length RLCG parameter matrix. BP neural network with strong nonlinear mapping ability is introduced to describe the random rotation of wiring harness to the ground. Combined with the finitedifference time-domain (FDTD) method, the crosstalk of the wiring harness is predicted. Experimental results show that the new method has good accuracy in predicting crosstalk of hand-assembled cable bundles. The higher the twisting degree of the wiring harness is, the more concentrated the crosstalk is.