Research Of Electromagnetic Inverse Problem Based On Adaptive Kriging Assisted Jaya Algorithm

The inverse problem of electromagnetic devices have always been the focus and difficulty in the field of electrical engineering in China.In the optimal process of electromagnetic devices,the efficiency and accuracy of the optimization should meet the application of practical engineering.The traditional optimization algorithm has the disadvantage of high computational cost.In order to overcome the disadvantage of computational cost,and improve the convergence speed and accuracy,the surrogate model was used instead of the traditional optimization algorithm.With the development of social technology,the combination of surrogate model and intelligent optimization algorithm has been widely used.Based on the blind Kriging surrogate model,which can greatly reduce the running times of finite element simulation software and improve the optimization efficiency..Meanwhile,combined with the Jaya algorithm for the optimization design of electromagnetic devices has become the research direction of this thesis.Firstly,In this thesis,the basic principle of the blind Kriging model is introduced in detail.Through the comparisions of different experimental design methods,the Latin hypercube design method is used to select the sampling points to build the blind Kriging model.The accuracy of blind Kriging model is tested and analyzed by using classical model evaluation indexes.Secondly,the standard Jaya algorithm is improved according to its current shortcomings.In the improved single-objective Jaya algorithm,Skew Tent chaotic mapping is added to initialize the population,and Levy Flight population updating strategy is used to update the population,thus the search space of the population is increased and the search ability of the global optimization algorithm is enhanced.Genetic algorithm（GA）,Particle swarm optimization（PSO）,standard Jaya algorithm and the improved single objective Jaya algorithm are compared and verified by some classical single-objective test functions.The optimal results show that the improved single objective Jaya algorithm has good optimal ability in single-objective optimization.In the multi-objective optimization algorithm,a multi-objective evolutionary algorithm based on decomposition（MOEAD）domain updating strategy is introduced to make the population converge to the Pareto frontier.Some classical multi-objective test functions are used to make comparative analysis of non-dominated sorting genetic algorithm-II（NSGAII）,MOEAD and improved multi-objective Jaya algorithms.The optimal results show that improved multi-objective Jaya algorithm has relatively uniform Pareto graph distribution and fast convergence speed in multi-objective optimization.Finally,the propose high effcient global optimization algorithm is applied to the single-objective and multi-objective optimization design for superconducting magnetic energy storage（SMES）and Loney’s solenoid,and the multi-objective optimization design for magnetic fluid hyperthermia system（MFH）and an interior permanent magnet synchronous motor（IPMSM）.The better results are obtained in the above application of the optimal design of electromagnetic devices.