OAK

Conceptual Design of Permanent Magnet Synchronous Motor using Topology Optimization Taehoon Jung School of Mechanical and Robotics Engineering Gwangju Institute of Science and Technology

Metadata Downloads
Abstract
본 논문은 영구자석 동기모터(Permanent Magnet Synchronous Motor, PMSM)의 성능 향상을 위한 개념설계를 제안한다. 첫째, 영구자석 동기모터의 전자기 토크 성능을 향상시키기위해할박배열(Halbach Array)을최적화하는다중재료위상최적화기반개 념설계를 제안한다. 둘째, 개념설계 연구를 소음 및 진동 저감을 위한 설계로 확장하여, 인공신경망(Artificial Neural Network, ANN)기반대리모델을이용한설계를제안한다. 본 논문에서 수행된 연구는 다음 세 가지 주요 내용으로 설명한다. 먼저, 할박 배열 최적화를 위한 다중 재료 위상최적화기반 개념설계를 제안한다. 할 박 배열은 영구자석을 특수하게 배치하여 자기장을 특정 방향으로 집중시키는 기술로, 전자기시스템의성능을향상시킬수있다.본논문에서는할박배열에대한자석조각의 형상,배치,자화방향을동시에최적화하기위해다중재료위상최적화를이용하였으며, 이를통해자기성능을극대화하였다.설계과정에서는물질보간법을이용하여영구자석 과철의형상을최적화하고,패널티기법을통해자석조각의자화방향을최적화하였다. 제안된설계방법은직사각형캐비티내부의자속밀도최대화와액추에이터의전자기력 최대화를 목표로 한 두 가지 설계 문제에 적용되었으며, 설계 유효성과 제작 가능성을 검증하였다. 다음으로, 제안된 다중 재료 위상최적화기반 개념설계를 자동차 인휠 모터(In-wheel Motor)에적용하였다.인휠모터는제한된공간에서높은토크밀도가요구되는구동모 터로, 본 논문에서는 평균 토크를 최대화하는 것을 설계 목표로 하였다. 토크 리플, 회전 속도,자화방향의개수를제약조건으로설정하였으며,설계결과를통해제안된방법의 설계 유연성을 검증하였다. 최적화 결과 모델과 초기 모델의 성능을 비교한 결과, 평균 토크는 증가하고 무게는 감소하여 토크 밀도가 향상되었다. 최적화된 할박 배열 인휠 모터는 상세 설계를 거쳐 제작되었으며, 역기전력과 토크 측정을 통해 성능 향상을 검증 하였다. 이를 통해, 제안된 다중 재료 위상 최적화 기반 개념 설계가 영구자석 동기모터 설계에 적용 가능함을 검증하였다. 마지막으로, 본 논문은 영구자석 동기모터의 개념설계 연구를 진동 및 소음 저감을 위한 설계로 확장하였다. 매입형 영구자석 동기모터(IPMSM)의 진동 및 소음을 저감하 기위한개념설계를위해인공신경망(Artificial Neural Network, ANN)기반대리모델을 이용하였다. ANN 기반 대리모델은 입력 데이터와 출력 데이터 간의 관계를 학습하며, 입력 데이터는 IPMSM의 고정자 구조 매개변수로 구성되고, 출력 데이터는 평균 토크, 토크 리플과 같은 전자기적 성능과 고유 진동수, 음압 수준(Sound Pressure Level, SPL) 으로구성한다.데이터는 ANN기반대리모델에서학습되며,학습된모델은성능요구를 충족하면서 다양한 구조를 빠르게 설계하고 평가할 수 있다. 본 논문에서는 전자기 성능 을 유지하면서 진동 및 소음을 저감하기 위해 ANN 기반 대리모델을 이용하여 IPMSM 고정자 구조를 설계하였다. 제안된 ANN 기반 대리모델을 이용한 개념설계는 서로 다른 목표 함수를 가진 두 설계 문제를 통해 설계 유효성을 검증하였다. 본 논문은 영구자석 동기모터의 전자기 성능과 소음 및 진동 성능을 향상시키는 최 적설계에 기여할 것으로 기대된다. ©2025 정 태 훈 ALL RIGHTS RESERVED|This dissertation proposes a conceptual design methodology to enhance the per- formance of Permanent Magnet Synchronous Motors (PMSMs). The study introduces two main approaches: first, a multi-material topology optimization-based conceptual design to optimize Halbach arrays for improving electromagnetic torque performance in PMSMs, and second, an Artificial Neural Network (ANN)-based surrogate model to extend conceptual design for reducing noise and vibration. The research conducted in this dissertation accounts for three topics. Firstly, a conceptual design based on multi-material topology optimization is pro- posed for the optimization of Halbach arrays. Halbach arrays, which enhance elec- tromagnetic system performance by concentrating magnetic flux in a specific direc- tion through the specialized arrangement of PM segments, should be optimized for the shape, arrangement, and magnetization direction of PM segments. The proposed methodology uses material interpolation to optimize the shape and arrangement of PM and iron, while a penalization scheme optimizes the magnetization direction of the PM segments. The design approach is applied to two design problems: maximizing mag- netic flux density within a rectangular cavity and maximizing electromagnetic force in an actuator. The effectiveness and manufacturability of the proposed design method were validated through the design and fabrication of two design problems. Secondly, the proposed multi-material topology optimization-based conceptual de- sign is applied to in-wheel motors for e-mobility. The in-wheel motor, which requires high torque density within limited space, is designed to maximize average torque while maintaining constraints on torque ripple, rotational speed, and the number of mag- netization directions. The results demonstrated improved torque density compared to the initial model, with increased average torque and reduced weight. The optimized Halbach array in-wheel motor is modified in the detailed design, and the prototype Halbach array in-wheel motor is fabricated. Performance validation through back-EMF and torque measurements demonstrated the applicability of the proposed conceptual design method to PMSM design. Lastly, this study extends the conceptual design to reduce noise and vibration in PMSMs. For this purpose, an ANN-based surrogate model was employed to design the stator structure of Interior Permanent Magnet Synchronous Motors (IPMSMs). The ANN-based surrogate model establishes the relationship between input data, defined as the stator structural parameters of IPMSMs, and output data, which includes elec- tromagnetic performance such as average torque and torque ripple, as well as noise and vibration characteristics like eigenfrequency and sound pressure level (SPL). The ANN- based surrogate model enables rapid evaluation and design of various stator structures while satisfying performance requirements. The effectiveness of the proposed concep- tual design using an ANN-based surrogate model was validated through two design problems with different objective functions, demonstrating the ability to maintain elec- tromagnetic performance while reducing vibration and noise. In conclusion, this dissertation contributes to the optimal design of PMSMs by en- hancing electromagnetic performance and reducing noise and vibration, offering prac- tical design methods for electric motor design. ©2025 Taehoon Jung ALL RIGHTS RESERVED – iii –
Author(s)
정태훈
Issued Date
2025
Type
Thesis
URI
https://scholar.gist.ac.kr/handle/local/19027
Alternative Author(s)
Jung Taehoon
Department
대학원 기계로봇공학부
Advisor
Lee, Jaewook
Table Of Contents
Abstract (English) i
Abstract (Korean) iv
List of Contents vii
List of Tables x
List of Figures xi
1 Introduction 1
1.1 Motivation and goal 1
1.2 Previous researches 5
1.2.1 Multi-material topology optimization of Halbach array 5
1.2.2 Design of PMSM with Halbach array 7
1.2.3 PMSM design for the reduction of noise and vibration 8
1.3 Research objective and contribution 10
1.4 Outline of dissertation 14
2 Multi-material topology optimization of PM segments 15
2.1 Introduction 15
2.2 Formulation 18
2.2.1 Governing equation 18
2.2.2 Multi-material interpolation scheme 19
2.2.3 PM fragmentation scheme 23
2.2.4 Non-magnetic frame structure using post-processing 25
2.2.5 Fabrication of Halbach array system 29
2.2.6 Optimization problem formulation 31
2.3 Design and fabrication example 33
2.3.1 Example.1 Rectangular cavity design 33
2.3.2 Example.2 Magnetic actuator design 38
2.4 Conclusion 44
– vii –
3 Performance analysis of permanent magnet synchronous motor 46
3.1 Introduction 46
3.2 Maxwell’s equation for magnetostatic analysis 48
3.3 Back-EMF calculation 50
3.4 Torque calculation 53
3.4.1 Torque profile calculation using maxwell stress tensor 54
3.4.2 Torque calculation using DQ-flux linkages 57
3.5 Loss calculation 62
3.6 Conclusion 63
4 Multi-material topology optimization of PMSM
with Halbach array 65
4.1 Introduction 65
4.2 Electromagnetic analysis 69
4.2.1 Magnetostatic analysis 69
4.2.2 Electromagnetic torque calculation 70
4.2.3 Base speed calculation 71
4.3 Multi-material topology optimization of in-wheel motor with Halbach
array 71
4.3.1 Specification of baseline in-wheel motor 72
4.3.2 Optimization problem formulation 72
4.3.3 Material interpolation scheme 75
4.3.4 Penalization scheme 78
4.4 Result and discussion 81
4.4.1 Pareto front solution between average torque and base speed 81
4.4.2 Effect of total numbers of discrete target magnetization direc-
tions on design result 83
4.4.3 Comparison with baseline motor 85
4.4.4 Fabrication and experimental validation 87
4.5 Conclusion 90
5 Conceptual design of PMSM for reduction of vibration and noise 92
5.1 Introduction 92
– viii –
5.2 Noise and vibration analysis of PMSM 95
5.2.1 Calculation of electromagnetic force 95
5.2.2 Modal analysis 97
5.2.3 Structure-acoustic analysis 98
5.3 Conceptual design of PMSM using ANN-based surrogate model 99
5.3.1 Surrogate model based on artificial neural network 100
5.3.2 Specification of initial IPMSM 103
5.3.3 Definition of design variables in PMSM 104
5.4 Result and discussion 106
5.5 Conclusion 114
6 Conclusion 116
6.1 Summary 116
6.2 Future works 119
6.2.1 Design optimization for loss reduction 119
6.2.2 Automation post-processing for detail design 120
6.2.3 Appling data sampling methods 120
References 122
– ix –
Degree
Doctor
Appears in Collections:
Department of Mechanical and Robotics Engineering > 4. Theses(Ph.D)
공개 및 라이선스
  • 공개 구분공개
파일 목록
  • 관련 파일이 존재하지 않습니다.

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