| In recent years,with the rapid update of wireless terminal communication technology and computer application design,many types of wireless terminal communication products have penetrated into people’s daily life,helping people to obtain and master various information in a timely and convenient manner.As the transceiver front-end of wireless terminal communication equipment,the overall performance of the antennas affects the quality of communication to a great extent.With the advent of the 5G era,wireless terminal communication equipments have put forward higher requirements in terms of miniaturization,multi-function,and high integration.At the same time,the number of communication bands that the equipment needs to support is gradually increasing,which makes the design and development of antennas facing many challenges.From the perspective of communication bandwidth,the antenna system should support as many frequency bands as possible to realize data transmission between different frequencies.The reserved design space of the antennas is also greatly reduced with the miniaturization and portability of terminal equipment simultaneously.Therefore,it is of great academic significance and engineering guidance value to study the multi-frequency and broadbandization of antennas in a narrow design space.From the spectrum point of view,bandwidth-intensive applications(such as video analysis,AIassisted computing,etc.)are widely favored,which means that the traditional crowded spectrum has been difficult to meet the high-quality,high-density communication environment.Millimeter Wave(mm Wave)band has a large amount of unused broadband spectrum,the channel is stable and the bit error rate is low,which can effectively solve the problems of highdensity data congestion transmission.However,the wavelength of the millimeter wave is extremely short,the path loss and air attenuation of electromagnetic waves during the transmission process are serious.How to complete the broadband and high-gain design of the antenna in the millimeter wave band has important research significance for the effective use of the millimeter wave.In addition,with the increasing update of integrated circuit technology,the data throughput and ultra-multi-threading processing capabilities of computers have been greatly improved.Various intelligent algorithms represented by machine learning or deep learning have achieved remarkable and successful applications in different fields,especially in the multi-source heterogeneous nonlinear information areas.In the field of electromagnetics,the use of machine learning to assist in the design of electromagnetic devices such as antennas and filters has been considered a powerful design method.Therefore,how to effectively combine the machine learning model with antenna modeling and optimization,supplemented by a mature model design process under the premise of meeting the communication needs,is still an urgent problem to be solved.In the meantime,this kind of interdisciplinary research is also the difficulty and hotspot of antenna design.Aiming at the above key questions,the thesis will focus on three aspects to carry out research and analysis.Firstly,the broadband antennas design is researched and discussed for the narrow design space of mobile terminal equipment,the problems of antenna multifrequency,broadband,mode distribution superposition,and bandwidth segment fusion are discussed in detail.Subsequently,the thesis will carry out detailed analysis and discussion on the application of various compact broadband high-gain slot antennas in the millimeter wave band,and integrate the array into the slot antennas.Finally,the thesis will use machine learning-assisted parameter optimization to analyze a metamaterial broadband antenna,and introduce the random forest model into the parameter optimization.The main working contents of this thesis are given bellow:1.This chapter illustrates the research background and significance of the broadband technology in wireless terminal communication,and expounds the machine learning methods in the electromagnetic fields.Combined with the topics of the thesis,the research status of broadband mobile terminal antennas,broadband millimeter-wave antennas,and machine learning optimization antenna design methods at home and abroad are analyzed and elaborated.These antennas and optimization algorithms are analyzed and discussed in detail with examples.The second section summarizing the working principles of monopole antenna,Yagi antenna and slot antenna,which provides a theoretical foundation for the following chapters.2.Broadband mobile terminal antennas.In this part,a seven-band mobile communication antenna with metal rim is proposed.With the help of double-strip structure and ground branches,the antenna has multi-mode characteristics.Besides,the antenna can achieve seven-band performance(824-960 MHz,1710-2690MHz)in a small design space with ground slot.Subsequently,on the basis of the aforementioned research,this chapter further designs an eight-band metal-rimmed broadband mobile terminal antenna.The paper analyzes and discusses the working mechanism of the antenna in the low and high frequency bands in detail.The proposed broadband antenna achieves effective coverage of the LTE700 frequency band within a narrow-reserved design space,and has the bandwidth extension capability of the high & low bands segmented design and step-by-step superposition of resonant modes,which solves the problem of unstable antenna resonance points and the high & low frequency bandwidths are affected by each other.Both antennas have good radiation efficiency and gain in the measurement.At the same time,this chapter also conducts a detailed theoretical analysis and experimental demonstration of the coaxial line’s errors on the antenna measured efficiency.3.Broadband,high gain mm Wave endfire antenna and array.Firstly,a millimeter-wave planar Yagi antenna unit loaded with parasitic elements is proposed.By introducing the double-sided parasitic elements around the director of the Yagi antenna,the gain of the Yagi antenna is effectively improved without affecting the physical parameters such as the size and profile,the impedance matching of the antenna is improved simultaneously.Substrate Integrated Waveguide(SIW)with symmetricafeil metal vias is employed as the feed and reflector structure in the proposed design,which further helps the Yagi antenna obtain a broadband bandwidth of dual-mode resonance.Based on the proposed Yagi antenna unit,a four-unit array is further designed in this part.The feed network is an oneto-four power divider composed of SIW,which makes the array have a compact size.Yagi antenna unit and array have demonstrated good broadband and high-gain end-fire radiation in the measurement.In the meanwhile,the substrate thickness of the two antennas is only0.25 mm,which also reduces the manufacturing cost for possible mass production.4.Broadband high-gain mm Wave one-sided directional slot antennas.First of all,a millimeter-wave slot antenna unit loaded with floating metal is designed.The floating metal is regarded as a combination of a special matching network and a reflector,which helps the slot antenna obtain the broadband of multi-mode resonance and partial onesided directional radiation.With the help of the coplanar transmission line,a compact dualunit slot antenna array is proposed later,which has high gain and good one-sided directional capability while maintaining broadband.Subsequently,this chapter explores the bandwidth and radiation properties of the proposed slot antenna unit/array with 3×3 and 4×3 circular elements AMC(Artificial Magnetic Conductor)reflectors loaded in the air layer.The unit and the array both keep compact structure,and show the one-sided directional radiation with broadband and high gain.5.Antenna broadband optimization method based on random forest model.In this section,a random forest model composed of some decision trees is designed and used to predict the geometric parameters of the antenna broadband design.In order to evaluate the prediction accuracy and generalization ability of the random forest model on this case,the thesis also introduces two other typical machine learning models(ridge regression and polynomial regression)for comparison.In addition,the thesis constructs and employs a smallsample training set to train three models,constructing a large-sample test set to verify and evaluate the accuracy & generalization ability of different models.Finally,the bandwidth optimization of the example antenna is completed with the help of a set of system modeling methods including the selection of the antenna parameter feature set,the cross-validation training and the generation of the optimal solution. |
PDF Full Text Request