Research On Multi-DOF And Multi-objective Rapid Optimization Method Based On Electromagnetic Simulation System

With the gradual development of microwave and millimeter wave devices in the direction of high performance,their boundaries and material structures have become more and more complicated.The traditional analytical method can only solve Maxwell's equations in a specific coordinate system and a regular shape,so the problems it can solve are very limited.Since then,researchers began to use computers and numerical analysis methods to perform numerical solutions.At present,the common method is to rely on electromagnetic simulation software and its internal local optimization algorithm to complete the ovrall design process of the device.However,due to the limited degree of freedom,this method can only carry out limited local optimization on the existing model.When the complex boundary of multiple degrees of freedom is introduced for design,it cannot be solved,so it cannot meet the high-performance design requirements of the device.In order to solve the above problems,this paper proposes a rapid design optimization method based on electromagnetic simulation system.It establishes a "surrogate model" of multi-parameter and multi-objective problems through the parametric modeling results of some design points,and introduces external global optimization algorithms to improve the accuracy of the solutions.Finally,through the data interface between the software,all the operations are concentrated in the same platform,so that this method can effectively simplify the design and optimization process of the device.The specific work completed in this paper are as follows:(1)In view of the large amount of simulation calculation and low efficiency of tradiditional design mode,this paper introduces the support vector regression machine model to construct the mapping relationship between input and output parameters,so as to replace the simulation software for fast calculation of the objective function value,and also through the HFSS-MATLAB-API toolbox realizes the function of co-simulation,so that all operations are concentrated in MATLAB,which reduces the threshold for users to operate.(2)The implementation principles of two types of intelligent optimization algorithms that are currently widely used in the electromagnetic field are introduced.In order to determine the applicable problem types of different algorithms so as to select the appropriate optimization algorithm in the example design,the sensitivity test of the paramaters of the two types of algorithms is carried out and the difference in the optimization performance of the algorithms is compared through the standard test function.(3)In order to verify the effectiveness of this rapid design optimization method based on electromagnetic simulation system,it is applied to the design of T-waveguide and coaxial-fed rectangular microstrip antenna.The first example completed the rapid design and optimization of T-waveguide,so that the T-waveguide realized the single-objective optimization requirement of power distribution at 10GHz.The second example introduces the rapid design and optimization process of the coaxial-fed rectangular microstrip antenna,which achieves rapid multi-objective optimization of the device.