| The Antenna,as a key component in the communication system,is responsible for transmitting and receiving signals,and its performance directly affects the communication quality.The iterative update of technology has brought about a demand for novel antennas,while the traditional design methods are unable to cope with the increasing complex antenna structures and high design freedom.In this paper,based on the new method of intelligent optimization algorithms and surrogate models,a fast multiparameter optimization design of microstrip antennas guided by performance requirements is realized.On the one hand,it is necessary to design an algorithm that converges fast and is not easy to trap into local optimum to quickly search for the antenna structure that meets our requirements in multidimensional design space.On the other hand,in order to further improve the efficiency and avoid large-scale full-wave electromagnetic simulations,it is necessary to study the construction of surrogate models with small samples.The main work includes the following aspects:(1)Proposed the improved whale optimization algorithm(IWOA).Four improved strategies are proposed to balance the two capabilities of local exploitation and global exploration of the algorithm to further improve the optimization ability.This is demonstrated by comparison experiments on 23 benchmark functions.The binary improved whale optimization algorithm(BIWOA)can solve optimization problems with binary variables by introducing transfer function processing.(2)The efficient optimization design of two kinds of microstrip antennas is carried out respectively based on IWOA and BIWOA.The bandwidth of ultra-wideband(UWB)antennas with rectangular cuts obtained in four structural parameters optimization experiments all cover 3.1 to 10.6GHz,and the return loss in the working band meets our requirements.The other is a frequency reconfigurable antenna with 24 switches,which can operate in different frequency bands with a low reflection coefficient in the resonance frequency by selecting switching states.(3)Based on the research described above,the number of electromagnetic simulations in the optimization process is further reduced by embedding surrogate models and the work efficiency is improved.Aiming at the difficulties of traditional surrogate model design,structure selection,threshold switching,and online updating strategies are proposed to construct the self-learning surrogate model.Based on the proposed algorithms and models,the optimization design of irregular polygonal UWB antennas and reconfigurable antennas with 12 switches are completed respectively,which significantly reduces the cost in the optimization process compared with the scheme without surrogate models.UWB antennas and reconfigurable antennas are the research hotspots in the field of modern communication and have a broad prospect of application.The scheme proposed in this paper solves some difficulties of traditional methods and successfully realizes the automatic optimization design of the above two kinds of antennas,which has reference significance for the intelligent design of other types of antennas. |
