CDEBP Neural Network And Researched On Its Application In Pre-assessments Of The Automotive Wiring Harness Crosstalk

In recent years,the technology of automobile development towards the direction of Electrification and intelligent. An increasing number of electronic products deterioration the electromagnetic environment of automobile. Automotive EMC problem is increasingly serious. Among which, the transmission line electromagnetic coupling problem is particularly prominent. Because the electromagnetic compatibility of automotive significantly impact vehicle safety and reliability, as well as the wire important position in the automotive electromagnetic compatibility problem. In this paper, on the basis of previous studies, Research on the pre-assessments of RTVU crosstalk of car wiring harness based on the theory of multi-conductor transmission lines (MTL). The key and difficult points on multi-conductor transmission lines (MTL) is how to obtain the per unit length distribution parameters of transmission line. In this paper, apply the method of BP neural network to forecast per unit length distribution parameters of transmission line. And apply the Cultural Differential Algorithm (CDE) to improve the performance of BP neural network. The author prime tasks are as follows:(1) Detailed derivation of the equations of multi-conductor transmission lines (MTL).Firstly, Introduce how to set up a uniform two-conductor transmission line Telegram equation (BLT equation). On this basis, derived inhomogeneous media multi-conductor transmission lines model and non-homogeneous media multi-conductor transmission lines model. And give a program flow chart of multi-conductor transmission line model.(2) Described the cultural differential evolution algorithm (CDE). First of all, introduce a source and development of cultural algorithm,the basic design framework and program flow chart of cultural algorithm and its characteristics. Then, Introduced the source,principle and the characteristics of the difference algorithm development. On this basis, embedded the differential evolution algorithm into the species space of culture algorithm, constitute a new kind of intelligent hybrid algorithm—Cultural differential evolution algorithm (CDE).(3)Apply the method of BP neural network to forecast per unit length distribution parameters of transmission line. The most accurate method to obtain the per unit length distribution parameters of transmission line is measured,However,it can not be obtained in the car design stage.And the method of theoretical formula calculate failed while used in the calculation the per unit length distribution parameters of transmission line of automotive wiring harness. So after detailed introduce BP neural network,We apply the method of BP neural network to forecast per unit length distribution parameters of transmission line. And detailed introduced principle of BP neural network and its advantages and shortcomings.(4)Cultural differential evolution algorithm applied to improve the performance of BP neural network. Combine Cultural differential evolution algorithm (CDE) and the BP neural network, and application of Differential Evolution Algorithm cultural training BP neural network weights and thresholds,to improve the BP neural network shortcomings and improve training efficiency. Then introduce the programming flow chart of cultural difference BP neural network.(5)Application of BP neural network based on cultural differential evolution algorithm (CDEBP) to forecast per unit length distribution parameters of transmission line.Firstly,set up training and forecast model based on CDEBP network to forecast per unit length distribution parameters of the transmission line.And details on how to set parameters of BP neural network architecture and relevant parameters of cultural differential evolution algorithm, as well as methods to achieved experimental samples. By analyzing the training and the forecast curve fitting curve, demonstrate that Application of CDEBP neural networks can forecast per unit length distribution of parameters transmission line very well. By analyzing the error convergence curve, demonstrate that Combine Cultural differential evolution algorithm (CDE) and the BP neural network improve the BP neural network shortcomings and improve training efficiency.(6)Take the wiring harness bind up Car reversing radar drive cable as an example. The per unit length distribution parameters of transmission line obtained by CDEBP neural network will be taken into the multi-conductor transmission line equation to pre-assessments of the automotive wiring harness crosstalk. And give the devices to measure conductor crosstalk. Comparison of crosstalk values have been predicted and measured to verify the method proposed in this paper can be used to pre-assessments crosstalk of vehicle wiring harness.,can be used to guide the car design pre-wiring job and improve vehicle electromagnetic environment.