Research On Scattering Properties Manipulation And Application Of Programmable Digital Metasurface

In recent years,electromagnetic metasurface has attracted great attention in scientific research because of its strong functional flexibility,low processing difficulty and strong potential application prospects.In fact,in the application design of electromagnetic metasurface,the construction and optimization of its element model is the key to realize electromagnetic performance.However,a lot of time and energy is spent on unit modeling,simulation and optimization in current designs.In this paper,based on the design of programmable electromagnetic metasurface,a deep learning-based method to explore the rapid design of unit modeling process and electromagnetic response optimization method is proposed,which can effectively solve the modeling and optimization of electromagnetic metasurface devices from the convolutional neural network(CNN).In this method,a pooled window for periodic coded metasurface design in checkerboard form is proposed,and thus,dimension reduction is carried out on the data to reduce the pool layer that is periodically added between the convolution layer as much as possible.Compared with the widely used traditional parameter scanning modeling and shallow neural network algorithm,the rapid element design method based on deep learning proposed in this paper can effectively solve the problem of high dimension and large amount of calculation of electromagnetic hypersurface element parameter data.The adoption of high dimensional microwave simulation modeling algorithm greatly reduces the time of array comprehensive manipulation and computational simulation of programmable electromagnetic metasurface,which plays a significant role in promoting the combined application of multi-functional metasurface units.As a validation of theoretical model,experiments are carried out to verify the scattering characteristics of radar cross-sectional area and low power communication in this paper.1.In this paper,a fast comprehensive calculation method of metasurface array based on deep learning is proposed.The electromagnetic response of a specific metasurface unit is simulated numerically,and the simulation results are used as training samples to train the integrated network model.Thus,the complex relationship between the electromagnetic response of the metasurface composed of the element and its array composition sequence and structural parameters is revealed.The structure parameters and coding sequence of the metasurface can be given adaptively within the set operating frequency,which reflects the advantages of this method for the design of RCS reduced metasurface.2.In this paper,by avoiding the errors caused by the coupling effect between the units in the design process,the intelligent algorithm is adopted to improve the design efficiency of the coding metasurface,and the optimization of the coding metasurface in the geometric parameters and spatial topology of the units is realized,which effectively improves the accuracy of the electromagnetic response of the units.The experiment results show that the coding metasurface is effective in the design and application of both a dispersive lens and a low RCS surface.3.In this paper,a wireless information transmission device based on programmable metasurface is proposed,which is different from the traditional communication system.The proposed transmitter based on the adjustable metasurface transmits information by modulating the reflection coefficient of the programmable metasurface.The receiver demodulates the information carried by it through echo detection.According to the experimental results,the transmission rate can reach 500 kbps,and the peak power of the adjustable metasurface transmitter is only 0.02 mw.In this paper,the rapid design and manipulation of coding metasurface scattering characteristics are taken as the target,and the problems of fast calculation from the electromagnetic response of the element to the scattering characteristics of the array of two-dimensional periodic metasurface are studied based on the hybrid deep neural network modeling technology.The accuracy of the proposed method for near-field and far-field control of electromagnetic waves was verified on holographic imaging surfaces,achromatic surfaces and RCS reduced surfaces.Based on the proposed method,a programmable electromagnetic metasurface transmitter was designed to realize wireless low-power transmission of information.This rapid design further advances the process of coding and programmable metasurface from theoretical design to application.