Antenna Optimization Design Integrating Improved Sparrow Search Algorithm And Neural Network Proxy Mode

An antenna is a device that converts electromagnetic waves into electrical currents or currents into electromagnetic waves,and is an integral part of wireless communication systems.In the era of booming wireless communication,the tasks undertaken by antennas are becoming increasingly heavy,and the requirements for antennas in systems are also constantly increasing.How to quickly design antennas that meet the requirements has become a work worth exploring.The traditional antenna design methods of manually adjusting parameters and using electromagnetic simulation software alone have problems such as large amount of calculation and time-consuming when designing multi-parameter antennas.At the same time,the system has higher and higher requirements on the performance of the antenna.When designing the antenna,it is necessary to integrate various performances and perform multi-objective optimization on the antenna.With the development of artificial intelligence,using surrogate model and swarm intelligence optimization algorithm to optimize antenna design has become an important research direction in antenna filed.Firstly,this paper constructs antenna surrogate models to replace traditional electromagnetic simulation software,which is used to solve the problems of large amount of calculation and time-consuming traditional methods.Four different methods are used to construct the antenna surrogate models,namely kriging,support vector regression,radial basis function,and BP neural network,and the performance of the surrogate models are analyzed.The experimental results show that the constructed BP neural network surrogate model can be used to predict the performance of the antenna,thus saving time and improving the efficiency of antenna design.Subsequently,in order to make the predicted value of the surrogate model closer to the real value,this paper uses the improved sparrow search algorithm to optimize the parameters of the model.In the late stage of solving complex optimization problems,the original sparrow search algorithm has slow convergence speed and reduced population diversity,which makes it difficult for the algorithm to achieve problem optimization.Therefore,this paper introduces three strategies of Bernoulli chaotic mapping,adaptive inertia weights and t-distribution to improve the original sparrow search algorithm,and proposes a multi-strategy improved sparrow search algorithm(MISSA).By comparing with other algorithms on 20 sets of test functions,the good performance of MISSA algorithm is proved.This algorithm is used to optimize the hyperparameters of the BP neural network model,construct the MISSA-BP surrogate model,improve the prediction accuracy of the model,and can be used to replace the electromagnetic simulation software.Finally,according to the multi-objective optimization requirement of multi-parameter tri-band microstrip antenna,this paper uses MISSA-BP surrogate model combined with multi-objective particle swarm optimization algorithm to carry out the multi-objective design of the antenna.The experimental results prove that the antenna designed using the method in this paper meets the requirements and takes less time.The obtained antenna samples are prepared into real objects,and the antenna data are measured.By comparing and analyzing the measured results with the simulation results,the effectiveness and practicality of the method proposed in this paper are further verified.