Neural Network Algorithm And Antenna Design Based On Particle Swarm Optimization

With the continuous development of communication technology,antennas are applied to various scenes.People have higher and higher requirements for antenna performance,and the traditional single frequency antenna can no longer meet the needs of development.At present,the antenna modeling method generally uses electromagnetic simulation software for modeling,but with the complexity of the model,the simulation time will be too long and the running time will occupy a lot of memory,which makes the efficiency of generating antenna model not high.The traditional electromagnetic simulation software has low efficiency in generating antenna models,and the use of single-frequency antennas has certain limitations and cannot be used in multiple communication frequency bands.This paper studies these two problems.Firstly,aiming at the efficiency of electromagnetic simulation,an antenna design model is built in this paper.The main function of the model is to input the performance parameters of the required antenna model and predict the size parameters of the antenna model.The antenna design model can be divided into two parts: the first part is the prediction of genius size parameters,this part uses the PSO algorithm to optimize the randomness of the initial weights and thresholds in the BP neural network to improve the model prediction antenna size accuracy;The second part is optimization,which is to further optimize the size parameters predicted by the neural network model,so that the corresponding performance parameters are closer to the set target value.In this paper,PSO algorithm is used for optimization,and the error formula between the required performance parameters and the performance parameters corresponding to the predicted size parameters is used as the fitness function,by continuously reducing the error,the optimal antenna size is obtained.Taking the microstrip rectangular antenna as an example,the mean square error(MSE)of the antenna size predicted by the neural network in this antenna design model is 0.0108,and the performance of the antenna optimized by the PSO algorithm is more in line with the set target value.Secondly,for the limitation of single frequency antenna,two dual frequency antenna models are designed by using the method of realizing multifrequency.The first one is to use multi branch technology to realize dual band antenna.The bandwidth of S11 <-10 d B optimized by HFSS simulation is 4 GHz-5.5 GHz and 7.8 GHz-8.7 GHz,and its relative bandwidth is 31.3%and 11.32% respectively,which can be applied to multiple communication bands.The second is the hexagonal prism helical microstrip slot antenna loaded with resonant ring.This model introduces a slotting technique and a loaded resonant ring structure,which makes the antenna become a dual-frequency antenna,and the gain is also improved.The bandwidths of the antenna S11 <-10 d B are 1.45GHz-1.78 GHz and2.39GHz-2.52 GHz,the corresponding relative bandwidths are 20.4% and 5.3%,and the maximum gain of this model is 3.54 d B.Its working frequency band can cover global positioning system(1.57542 GHz),Bluetooth(2.4 GHz-2.5GHz),WLAN(2.4 GHz-2.484 GHz),Wi Fi(2.4 GHz-2.484 GHz)and other communication frequency bands.Finally,this paper combines the antenna design model with the above two dual band antenna models to verify the feasibility and accuracy of using the antenna design model method,and compares it with the traditional electromagnetic simulation software.By comparing the traditional electromagnetic simulation software with the antenna design method proposed in this paper,the new method requires much less computation,saves time,and has a better overall effect.