| Microwave filters are the essential devices of the transmitter and receiverin modern communication systems as they can separate the signal. It makes usefulsignal pass by as no attenuation as possible and the unwanted signal attenuation aslarge as possible inhibition of its adoption. With the development of wirelesscommunication technology, the filters must perform more reliability as increasinglynarrow band between the signals. Coaxial cavity filters have the advantages of highQ, low insertion loss, high power capacity, easy to manufacture and its stableperformance, so, they have a high application value. Therefore, it is necessary toresearch the optimation methods of coaxical filers design.The optimation design methods of cavity filters are mainly studied in this paper.Firstly, cross-coupling matrix synthesis and transformation using Matlab programare studied; Then, TE01mode dielectric resonator filters have been studied and amethod to suppress the high-order mode is proposed. Finally, this paper studies themethod that using neural network to optimize cavity filter design. The mijor workdone as follows:1. First, briefly reviewed the basic principles of microwave design.Respectively,introduced the Chebyshev low-pass filter prototype as well as frequency conversionand inversion transform.2. Mainly studied the cross-coupling theory and an effective solution wasprovides by using Matlab program to sythesizt and similarity transform couplingmatrix. In order to test the feasibility of the Matlab program, an example thatcoupling matrix synthesis and simplification using Matlab program is given.Simulation results calculatede by the Matlab program was fully agreed to theequivalent circuit simulation curve given by AWR. So, it proved that the Matlabprogram can be used in filter design.3. Studied the design of TE01mold dielectric cavity filter and a method forsuppressing high-order mode by adjusting the ratio of dielectric resonator wasproposed. First, this paper analyzed TE01mode dielectric resonator and studied itselectromagnetic field distribution characteristics.The method that by adjusting the ratios of the various parts of dielectric resonators to suppress higher-order mode wasproposed.Then, the coupling structure between dielectric resonator was studied as itwas different form metal resonators.Positive coupling or negative couplingbetween dielectric resonators are mainly decided by their electromagnetic fielddistribution. Finally, a band-pass dielectric cavity that can be used in LTEcommunication is given.Test results showed that the cavity dielectric filter not onlyhad good high-mode suppression but also had very low insertion loss.4. Studied application of neural networks in the process of microwave filterdesign and proposed by establishing neural network model to extract couplingparameters of microwave filters.Using neural network to design filter can avoidrepeating the simulation process, improve filter design efficiency and short filterdesign cycle. Compared to traditional filter design methods, neural network moelingtechniques had many advantages, so using neural network to optimize filter designnot only can short the dising cycle but also can improve filter design accuracy. Inthis paper, studied how to use neural network to optimize filter design.And then, aminiaturization coaxial cavity band-pass filter is designed in neural networkoptimization method by using λ/4-type SIR as the basic resonance unit.Simulationresults show that the step-impedance resonator cavity filter not only can meet thedesign requirements but also can effectively reduce the filter size. At the same time,it also proved that neural network can be used in filter design. |
PDF Full Text Request