Research On Artificial Intelligence Based Rapid Simulation And Optimization Method For Antenna Design

Antenna,as an important device for converting electrical signals and electromagnetic signals at the end of wireless communication systems,its performance significantly affects the quality of information transmission.The demand of broadband antennas becomes more and more pressing in modern communication systems.However,the art of state of design method of antenna faces the problems of process redundancy and low efficiency in manually optimizing antenna performance.With the rapid development of artificial intelligence,the agent model is used for predicting the highly non-linear relationship between antenna structure and performance,in order to reduce the number of electromagnetic simulation software calls.Achieving rapid simulation and optimization design as well as improving the simulation speed and optimization efficiency through the combination of the agent model and intelligent optimization algorithms will become an important method to solve the inefficiency of artificially optimized antennas.This dissertation uses artificial intelligence to achieve the optimized design of the Yagi antenna,then breaks the conventional antenna optimization thinking,an undefined structure planar antenna with discretized matrix elements is used as an optimization object,and finally a software based on proposed algorithms is designed and developed for antenna design and optimization application.This dissertation completed algorithm innovation and engineering application.The specific research content is divided into three parts:(1)An optimization design scheme is proposed by combining XGBoost regression agent model and PSO algorithm,for optimizing the structural parameters of the Yagi antenna.The data is simulates based on the structural characteristics of the Yagi antenna,and then the simulated data is used to train the XGBoost regression agent model.The average mean square error of the XGBoost regression agent model is 3.20,which is lower than homogeneous regression agent models.Next,the optimization goals and important parameters of the PSO algorithm are designed.Finally,based on the designed XGBoost regression agent model and PSO algorithm,the structural parameters of the Yagi antenna are optimized.The optimized Yagi antenna's performance fully meets the predetermined requirements,whose return loss at 10.29GHz reaches-49.003dB.This antenna has broadband characteristics with a relative bandwidth of 20.12%in 9.65GHz-11.72GHz.(2)An optimization design solution for the undefined structure planar antenna is proposed based on the XGBoost classification agent model and the VPSO algorithm,in which the artificial intelligence method is used to design the undefined structure planar antenna with discretized matrix elements.In the scheme,the antenna patches are discretely divided into 8*7 square cells by grid division with two states of 0/1 in each cell.The XGBoost classification agent model is trained based on simulation data of the antenna model.The accuracy of the XGBoost classification agent model is 92.90%,which is higher than homogeneous classification agent models.The VPSO algorithm updates the position conversion function of PSO,and then sets optimization objective function.Finally,based on the completed XGBoost classification agent model and the VPSO algorithm,the structure optimization of the planar antenna achieves the predetermined optimization target.The experimental optimization results show that the antenna structure automatically generated by the method fully meets the predetermined demand.The antenna return loss reaches-32.46dB at the 5.44GHz.The antenna working band is 4.11 GHz-5.80GHz with a relative bandwidth of 31.07%,which meets the requirements of broadband communication.(3)This research puts forward an antenna optimization design platform,which includes the front-end display and the background services.The front-end display consists of the parameter setting,file management and optimization display modules.Next,the background services is composed by the optimization algorithms,model training and co-simulation modules.The software platform achieves the visual display of the antenna optimization through the artificial intelligence methods,as well as presents the antenna optimization process and functions more clearly.The optimization results of the Yagi antenna and the undefined structure planar antenna show that the proposed scheme has excellent performance in the intelligent design and efficient optimization of the antenna.These schemes provide a cutting-edge and practical reference for the intelligent optimization design of the antenna.