Optimum Design Of Winding Conductors Transposition Of Power Transformer Based On Universal Kriging Model

The synthesis research of engineering electromagnetic field can be divided into two aspects,namely the positive and inverse problems of engineering electromagnetic field.In practical engineering,the optimization of electromagnetic devices is often required,that is,the inverse problem of electromagnetic fields.In the other word,the electromagnetic inverse problem is to achieve the ideal parameters or performances of electromagnetic devices through optimal design of these devices.At present,some optimization algorithms have been studied for solving the electromagnetic inverse problem including genetic algorithm,particle swarm optimization algorithm,ant colony optimization algorithm,neural network optimization algorithm and simulated annealing algorithm and so on.However,when these optimization algorithms are applied to solve the complex engineering problems,the convergence speed and the accuracy of algorithms is difficult to meet the numerical calculation of electromagnetic inverse problem.In order to find the optimal winding transposition of a power transformaer,in this paper,a novel hybrid global optimization algorithm will be proposed by combining particle swarm algorithm and universal Kriging model.This proposed algorithm is different from the traditional stochastic optimization algorithm,of which a numerical calculation of electromagnetic field for each iteration will be required.It is a more accurate response model construted by semi-parameterization.Meanwhile,some disadvantage can be solved for the complex electromagnetic inverse problem,such as huge computing time,low convergence speed and large computer memory and so on.The detailed disscussions of this paper will be given as follows:First of all,after learning the global optimization algorithm based on the Kriging model,according to the requirements of high precision of Kriging model,a novel,efficient sampling technology will be researched,so that the proposed universal Kriging has a high accuracy with the minimal sampling points.The Kriging models are programmed by the Matlab.Then,based on the Kriging interpolation technique,a more accurate universal Kriging will be studied,and a more accurate efficient global optimization,through the step by step iterative way,is proposed,which is combined with the particle swarm optimization algorithm.A computer software for the proposed optimization algorithm is programmed by the C++.Finally,according to the characteristics of single-phase power transformer,a two-dimensional simulation model is constructed by the Maxwell software.Meanwhile,due to the characteristics of winding transposition,some design variables for reducing circulating current loss are given and optimized by the proposed algorithm.After the optimization,the circulating current loss is greatly reduced,and the effectiveness of the proposed algorithm is verified.