Research On Electromagnetism-like Mechanism Algorithm And Its Application In Electromagnetics

Electromagnetism-like mechanism(EM)algorithm,a brand-new intelligent optimization algorithm proposed by S.I.Birbil and S.C.Fang in 2003,is inspired by the attraction-repulsion mechanism in the electromagnetic field.EM algorithm has the advantages as simple optimization mechanism and high convergence rate.Compared with other algorithms,EM algorithm shows a very strong global search capability.However,it performs poor convergence in the settlement of part of high-dimensional or multimodal problem,which means it still easy to fall into a local optimal value.Electromagnetic synthesis problem is always complex mathematical problem with complex model structure and nonlinear objective function.As a result,many traditional optimization algorithms no longer meet the requirements of high accuracy and efficiency of electromagnetic synthesis.Multi-objective problems usually contain multiple objective functions,corresponding to different design indicators.Weighted aggregation(WA),turning the multi-objective problem into a single-objective problem,is one of traditional multi-objective methods.However,it is difficult to determine the combination of the weighting coefficients for each sub-item,which is likely to cause only partial indices to meet the requirements,but the remaining unsatisfactory.It means the design process could be tedious and time-consuming because designers need to try again and again.To solve above two problems,an improved EM algorithm is proposed in this paper,which is applied to two single-objective electromagnetic problems.And then,a multi-level optimization(MLO)method based on the improved EM algorithm is presented,which is applied to three multi-objective electromagnetic problems.The specific works are listed as follows:(1)The EM algorithm is briefly introduced from three aspects: the basic principle of the algorithm,the algorithm description and the algorithm flow,which lays a theoretical foundation for the improvement of EM algorithm.(2)An improved EM(IEM)algorithm is proposed.In the IEM algorithm,mutation operator is adopted by selecting particles with certain small probability to improve population diversity.Normal benchmark functions and classical high-dimensional benchmark functions were used to test and analyze its performance.And the simulation results of new algorithm is compared with EM algorithm,DE algorithm and PSO algorithm.(3)The IEM algorithm is applied to the single-objective electromagnetic problem synthesis.First,applying IEM algorithm to FIR digital filter synthesis,including low pass FIR digital filter and band pass FIR digital filter.The simulation results are compared with those of EM algorithm,GA algorithm and PSO algorithm.And then,applying IEM algorithm to double-layer square FSS,which is used as the superstrate of microstrip antenna.The simulation results for the directivity and gain of the new antenna is compared with those of the original antenna.(4)The IEM algorithm is applied to the multi-objective electromagnetic problem synthesis.A method for multi-objective optimization is proposed,in which MLO method is introduced.The method incorporates a simple concept whereby multiple sub-objective functions,representing different requirements in multiple objective problems,are minimized one by one in the IEM algorithm.Applying the proposed method to microwave devices synthesis,including two L-shaped folded monopole antennas,a compact dual band slot antenna for WiMAX/WLAN applications and a dielectric filter.The simulation results are compared with those of existing literature.