Research On Crosstalk Of Multi-conductor Transmission Line Based On FDTD Algorithm

Electric energy is a necessity for the production and living of society.A large number of electric equipments achieve a normal work on the condition that the circuits are energized by electric energy.Transmission lines are the basic component to connect electrical equipment.The more functions and components of power equipment,the more transmission lines are needed.So a large number of transmission lines with different functions will be closely linked.In the low-frequency band,the selection of transmission line types can meet the requirements only if the current carrying capacity is considered.However,the frequency raising is inevitable in order that the devices can be made smaller and lighter.The transmission lines are not only an electrical connecting element at high frequency band,but also a plenty of electromagnetic interference(EMI)source and susceptive device.The influence of equipments can not be ignored.In order to control or weaken the electromagnetic effect and realize the electromagnetic compatibility(EMC)of various equipments and components,the crosstalk must be calculated accurately.The main contents are as follows.Firstly,a variety of transmission line models are studied in this thesis.The characteristics of the transmission line parameter matrix and physical significance of the model of the MTLs are studied in this thesis.The parameter matrix of parallel MTLs is extracted by analytical method and numerical method,respectively.On the basis of the telegraph equation and the parameter matrix,the methods for solving crosstalk of the MTLs are introduced from time-and frequency-domain,respectively.The algorithm of the conventional ? type leapfrog difference format,the new implicit Wendroff difference format and the Kambiz Afroz difference format are analyzed in detail in the time-domain while the modulus decoupling method is analyzed in the frequency domain.Secondly,the crosstalk of multi-core twisted cable is studied.And a method extracting the parameter matrix based on the BP algorithm is proposed in this thesis.With properly selecting the parameter matrix samples of the multi-core twisted cable,the BP neural network algorithm based on the Levenberg Marquardt algorithm can easily learn the change characteristics of the parameter matrix.The crosstalk of multi-core twisted cable can be predicted by the new method combined with the FDTD algorithm.Taking the three-core twisted cable as an example,the accuracy of the new method tested in the frequency band of 100 k Hz-1 GHz is verified,and the results show that the proposed method has high accuracy and performance for the crosstalk of multi-core twisted cable.Thirdly,a new method with high efficiency for solving the crosstalk of random wiring harness is proposed in the studying of the crosstalk of random wiring harness.The random harness based on the cascade transmission lines theory is modeled by pseudo-random numbers.The various factors considered in the modeling include the random exchange of the wires position in the harness section and the rotation of the harness section to the ground.The determinant transformation of the parameter matrix is used to solve the problem of random exchange of wires,and BP neural network is used to deal with the problem of rotation.The crosstalk of random harness can be also predicted combined with the FDTD algorithm.Compared with the experiment and the simulation,it can be seen that the new method has a better prediction ability for the crosstalk of random wiring harness.