Please use this identifier to cite or link to this item: http://hdl.handle.net/10397/35468
Title: Multi-objective optimal design of electric motor and controller integrated systems based on parameterized FEM
Authors: Yang, Lin
Advisors: Ho, S. L. (EE)
Fu, W. N. (EE)
Keywords: Electric controllers -- Design and construction.
Electric motors -- Design and construction.
Issue Date: 2015
Publisher: The Hong Kong Polytechnic University
Abstract: In this thesis, fast optimal design of electric motor and controller integrated systems is investigated based on parameterized finite element method (FEM) coupled with global optimization methods. Novel methods both for single objective and multi-objective problems will be reported for fast optimization on applications of motor design. For an accurate analysis of the electromagnetic field, a parameterized FEM is applied considering the nonlinear handling, slave master technique and circuit coupling technique. To speed up the calculation of FEM, mesh techniques are introduced and applied including the parameterized mesh generation and refinement methods, adaptive DoFs FEM and remesh-free method. To consider the motor optimization in a system level, the pulse width modulation (PWM) techniques are introduced and discussed particularly in the context of the motor optimization design. Switching losses and harmonic effect are considered in the performance evaluation of the EMs. Powerful optimization methods are then proposed both for single objective and multi-objective problems. An FEM based sensitivity analysis approach has also been proposed to further speed up the optimization process. The hybrid optimization methods are then combined with the parameterized finite element method and well applied to improve the performance of electrical devices such as magnetic gears and controller integrated electric motors.
In this thesis, the following work has been done: (1) 2D FEM programs based on A-{482} potential formulations have been implemented for solving general field problems. Techniques for nonlinear material, movement and circuit coupling are also discussed. (2) 2D parameterized mesh methods are studied for fast FEM modeling. A remesh-free mesh deformation method (both in 2D and 3D) is proposed to reduce the EM modeling time neither re-generates the mesh nor increases the number of unknowns. Updated meshes can be directly derived using a coordinate mapping technique. (3) Discussions of machine properties and applications related to variable-speed have been introduced. General open loop control technique and SPWM technique are introduced and coupled with the FEM calculation of the motors. Therefore optimization designs can be performed in the system level by considering the performance of controller. (4) A novel direct sensitivity analysis approach based on FEM has been introduced, which can be widely used in practical optimization problems. The sensitivity information can be obtained by only post-processing the solution from the FEM calculation. (5) A multilevel hybrid numerical optimization method has been employed to optimize the performance of devices we considered. A multi-level method has been proposed firstly, where parameters will be defined with different levels according to the sensitivity analysis. The optimal parameters with lower levels, which mean a less important impact on the motor, will be assigned a low accuracy while the parameters in the high level with high accuracy. (6) Hybrid stochastic optimization methods combined with the sensitivity information have been applied in motor optimization. The sensitivity is defined as a factor to guide the local search in the global optimization. Both merits of these two kinds of optimal methods mentioned are utilized to speed up the motor optimization. A great reduction of the computational burden is reported. (7) Multi-objective optimal designs have been implemented to help designers to make better decisions according to the performance of the motors. An optimal metric is proposed to quantitate the improvements in both the whole objective functions and specific objectives. Multiple probability vectors are applied to improve the diversity of the solutions in multi-objective applications. The major contributions in this thesis are summarized as follows: A novel remesh-free method is proposed and combined with the parameterized mesh techniques to reduce the mesh regeneration time during the optimization process. A novel sensitivity analysis approach based on parameterized FEM has been proposed to benefit the pertinence analysis of design parameters and guide the local search in the sub-process of the global optimization. The optimal design of EM is considered in a system level by integrating the controller. Harmonic effects and losses due to the controller are considered within the performance evaluation of the EMs. An improved Tabu Search Algorithm and a novel self-adaptation evolution strategy are proposed to speed up the overall optimization procedure. The merits of stochastic and deterministic search methods are both incorporated by combining the sensitivity factor. Motor designs are investigated and evaluated within the multi-objective context, where a clear relationship among different objectives interested will be given, and preferred compromise solutions can be chosen for further study.
Description: PolyU Library Call No.: [THS] LG51 .H577P EE 2015 Yang
xxii, 188 pages :illustrations (some color)
URI: http://hdl.handle.net/10397/35468
Rights: All rights reserved.
Appears in Collections:Thesis

Files in This Item:
File Description SizeFormat 
b28238047_link.htmFor PolyU Users203 BHTMLView/Open
b28238047_ir.pdfFor All Users (Non-printable)7.35 MBAdobe PDFView/Open
Show full item record

Page view(s)

94
Last Week
4
Last month
Checked on Sep 18, 2017

Download(s)

45
Checked on Sep 18, 2017

Google ScholarTM

Check



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