Please use this identifier to cite or link to this item:
Title: Finite element analysis with embedded global optimization method for optimal design of electric devices
Authors: Chen, Ningning
Keywords: Finite element method.
Electromagnetic devices -- Design.
Hong Kong Polytechnic University -- Dissertations
Issue Date: 2012
Publisher: The Hong Kong Polytechnic University
Abstract: In this project, three-dimensional (3D) Finite element method (FEM) embedded global optimization method is applied to obtain optimal design electric devices. FEM with nodal basis function and edge basis function are used to analysis magnetic field and the eddy current problem in electric devices. Time step FEM with slave master technique, circuit coupling technique is applied to simulate the performance of electric motor. Global optimization method including Genetic algorithm (GA), simulated annealing (SA), taboo search (TA) and particle swarm optimization (PSO) are coupled with FEM program to find the parameters of the optimal design. Parameter extraction technique is applied to extract the mass parameters of electric motor to accelerate the computation speed of optimization process. Moving least square (MLS) based surface response model is applied to reduce the optimization time. The coupled FEM with optimization method is applied to optimize the performance of surface mounted PM motor, magnetic gear and an axial flux magnetic motor. In the thesis, the following work has been done: (1) 2D and 3D FEM program for eddy current have been implemented and couple with optimization method. (2) A two-grid FEM has been studied for reducing the 3-D FEM computation time to reduce the computation time of nonlinear problems. TEAM Workshop problem 13 is used to test the two-grid method and the results obtained using the two-grid algorithm are compared with those using conventional methods. Since the two-grid method requires less computing time, it can be effectively applied to study large-scale nonlinear problems. (3) Particle swam optimization (PSO) optimal algorithm and genetic algorithm (GA) have been implemented for optimization computation. A surrogate based on moving least squares (MLS) method has been developed to approximate the expected fitness evaluations, which would replace many times of FEA computation and would save much computation time. FEM with embedded GA has been applied to optimize the shape design of permanent magnet (PM) in motors to reduce the cogging torque. (4) Moving mesh FEM has been proposed for coupling FEM with optimization method. (5) The state of art programming method with object oriented programming technique for FEM is introduce to faculae the complexity of programming. The major contributions of the thesis are reflected in the following aspects: It proposes a novel moving mesh method to handle the mesh re-generation in the optimization step; Two-grid method is introduced to reduce the computational time of the nonlinear Maxwell system with nonlinear materials, and an effective interpolation of the FE solution from coarse grid to non-nested fine grid is proposed. 3D nodal and edge FEMs are implemented to analyze the eddy current problem involved in electric devices, and coupled with some global optimization process to design optimal electric devices. The FEM with global optimization method is applied to optimize the performance of PM motors and magnetic gears. Slave master technique is extended for 3D FEM computations to handle the non-conforming meshed on the interface between the moving and static parts of the devices. Time step FEM along with the slave master technique and circuit coupling technique is used to simulate the performance of the electric motors.
Description: xii, 131 leaves : ill. ; 30 cm.
PolyU Library Call No.: [THS] LG51 .H577P EE 2012 Chen
Rights: All rights reserved.
Appears in Collections:Thesis

Files in This Item:
File Description SizeFormat 
b25512511_link.htmFor PolyU Users 162 BHTMLView/Open
b25512511_ir.pdfFor All Users (Non-printable) 3.33 MBAdobe PDFView/Open
Show full item record
PIRA download icon_1.1View/Download Contents

Page view(s)

Last Week
Last month
Citations as of Sep 17, 2018


Citations as of Sep 17, 2018

Google ScholarTM


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.