Optimal Design Method Based On Extreme Learning Machine And Its Application To Inverse Electromagnetic Field Problem

Electrical equipment is an extremely important part of the power system.With the increasingly fierce market competition,the optimization design of electrical equipment has been paid more and more attention by scholars.Through the optimization design of electrical equipment,production cost can be reduced,operation performance can be improved,operation loss can be reduced,with obvious economic benefits.However,nowadays,optimization design problems are increasingly complex,presenting characteristics of multi-variables,multi-peaks and multi-constraints.The combination of traditional finite element analysis and global class optimization design method can no longer meet the requirements of rapid and efficient optimization design due to its complex process and long calculation time.Therefore,in this context,it is of great significance to study a more adaptable optimization design method.To solve the above problems,this paper studies a new surrogate model based on artificial intelligence named Extreme Learning Machine(ELM),and combines it with optimization algorithm to construct an optimization design method with better performance.Firstly,this paper studies the basic theory and algorithm flow of ELM,and constructs the surrogate model by using the prediction function of ELM.Compared with other surrogate models,the fitting and generalization performance of ELM is verified.At the same time,in order to improve the accuracy of the surrogate model constructed by ELM,Adaptive-Sampling Kriging Algorithm is studied in this paper.This method ensures the randomness,accuracy and diversity of sampling by combining three sampling steps: Multi-Jittered Sampling,Compact Search Sampling and Exclusive Space-Filling Method.Secondly,the extreme learning machine algorithm and particle swarm optimization algorithm are combined to construct the ELM-PSO.This paper studies its mathematical model and algorithm steps in detail.In order to verify the performance of ELM-PSO,this paper selects several kinds of standard test functions for experimental analysis,and verifies the excellent optimization ability of ELM-PSO by comparing the results with other optimization design algorithms.In engineering test,the design problem of superconducting magnetic energy storage is selected to further test the performance of the algorithm.Finally,the ELM-PSO method studied in this paper is applied to the optimization design of electrical equipment.Firstly,a surface-mounted permanent magnet synchronous motor is taken as the research object.Through the simulation of its electromagnetic field and temperature field and their coupling,the performance of the motor is analyzed.This paper uses the optimization design algorithm to optimize the motor and obtains the optimal design scheme.Through optimization,the optimal design of permanent magnet synchronous motor loss is realized,and its temperature field is analyzed.In addition,the torque ripple of a large power Vtype interior permanent magnet synchronous motor is optimized by using the algorithm to improve the motor performance.