| Microstrip antennas are widely used in the field of MEMS.However,when designing a microstrip antenna,it is usually necessary to calculate the theoretical solution of the antenna size with the help of formulas.However,there is a certain deviation between the theoretical solution and the actual size of the antenna,and parameter scanning analysis is required to obtain the optimal antenna size.Therefore,this paper proposes a new idea to design a microstrip antenna,Based on the actual antenna size data set training model,the optimal antenna size value can be obtained quickly by using the cyclic neural network and RPSO-ELM network algorithm,which will help the simulation personnel get familiar with the antenna characteristics in advance,improve the simulation efficiency,and can also be used to verify the proposed new antenna model.In the process of microstrip antenna design and simulation,a lot of formula calculations are often required,so this paper attempts a new method and means,using the cyclic neural network and RPSO-ELM network algorithm to train the model,and use it to determine the size value of rectangular patch antenna when the dielectric constant and resonant frequency are given.This paper compares several different algorithms,using antenna data,of which 90 groups of data are used for training and 10 groups of data are used for testing.Finally,RPSO-ELM network is selected to build and tune the model.At the end of this paper,a rectangular patch microstrip antenna is designed using HFSS software,and it is simulated and tested.The test results show that the designed antenna has good performance.Then this paper uses the dielectric constant and other eigenvalues of the designed antenna to predict in the RPSO-ELM network model.Finally,the actual size of the antenna is compared with the predicted size value obtained from the model,Through comparison,it is found that the error of the model prediction is small and the accuracy is high.The results show that the size value of the microstrip antenna can be obtained efficiently without theoretical formula calculation using the algorithm model,which greatly reduces the complexity of the antenna design simulation process.Moreover,the trained model is simple and easy to use,and can be used in the antenna design and simulation process.The algorithm model can be further encapsulated into a size recommendation system for optimal design of rectangular patch microstrip antenna.As long as the antenna eigenvalue is assigned to the system,the system model will automatically and accurately obtain the size value of this antenna,thus simplifying the process of antenna optimal design. |