Study On The Optimization Method Of Space Mapping Algorithm In Antenna And Microwave Devices

It is time-consuming to use the full-wave numerical analysis method to describe the performances of antennas or microwave devices whose structures are quite complex.In the process of optimization,full-wave electromagnetic simulations are needed repeatedly to evaluate its performance.But the full wave simulation is time-and memory-consuming,which makes it infeasible for efficient optimization.Therefore,how to use fewer high precision full-wave electromagnetic simulations to quickly get optimal design parameters,become the key technology of complex electromagnetic device optimization problem.Among them,using a variety of surrogate models to build the mapping relationship between design variables and objective function,to avoid a large number of high-precision fitness function evaluation,has become the efficient way to solve this optimization problem.In this paper,we use space mapping and manifold mapping methods to solve above problem.An efficient optimization of antenna array,substrate integrated waveguide filters and lump-loaded antennas are carried out.Space mapping is different from traditional electromagnetic optimization.It divides the structures requiring optimization into two models:a coarse model with low simulation precision and less time-consuming;a fine model with high simulation precision but more time-consuming.By establishing the mapping relationship between the two models,the optimization is carried out in less time-consuming coarse model space and the verification in the fine model to improve the optimization efficiency without sacrificing precision.Manifold mapping algorithm,an improvement on the space mapping algorithm,replaces the coarse model with fast computing speed and low precision by constructing an surrogate model.The surrogate model adopts the mathematical manifold idea and includes both coarse and fine model responses and directly establishes the mapping relationship with the fine model in the surrogate one,making the response between surrogate model and the fine model closer.Thus,compared to space mapping algorithm,this can simplify the optimization by reducing additional parameter extraction.The main contribution of this dissertation includes:Firstly,the space mapping algorithm is applied to the optimization of antenna arrays.It solves the time-and labor-consuming shortcoming of traditional antenna array optimization,which efficiently optimizes the antenna array scattering and radiation.The example of reduction radar cross section of antenna array and shaped beam for low profile antenna array of Beidou navigation system star indicates the fast speed and high efficiency of space mapping in the optimization of antenna array.Secondly,the space mapping algorithm is applied to the design of substrate integrated waveguide filters.The substrate integrated waveguide,due to two rows of metallic periodic holes of its structure features,increases the unknown quantity,as well as the computation time and memory consumption in full wave analysis.In order to realize fast and efficient optimization,the application of space mapping by combining equivalent circuit and the full wave analysis can optimize the design of substrate integrated waveguide band-pass and high-pass filter in favor of improving the stop-band performance.At last,the method of optimization of lump-loaded antennas by manifold mapping algorithm with the method of moments accelerated by adaptive cross cpproximation(ACA)is proposed.In the process of optimization,the coarse model has no empirical formula and the equivalent circuit is not easy to simulate.Thus,the accuracy of the model is precisely controlled by the method of moment of adaptive cross approximation algorithm acceleration.The optimization models of coarse model and the fine model are constructed by controlling the precision of adaptive cross approximation algorithm.In addition,the manifold mapping algorithm overcomes the shortcomings of spatial mapping method and the additional parameter extraction of the calculation of the mapping function,thus improving the efficiency of optimization.