隨著科技的發展，馬達與日常生活密不可分，一般系統廠中僅以現有馬達規格透過試誤法進行簡易參數調整，以符合不同客戶的規格與特性需求，其主要受限於過長設計時間與開發人力缺乏，重新設計在模擬軟體內需要耗費許多時間以及精力，同時針對不同性能所需要設計參數也需要專業能力與經驗。
為了解決上述問題，本研究針對現行馬達設計之需求，以內轉子式表面貼磁馬達(Surface mounted Permanent Magnet Motors, SPMMs)之操作設定開發自動化流程腳本平台，同時將每個功能模組化，不但能大幅減低操作時的不便，未來更能透過此模組化平台，將自動化腳本針對不同場景的馬達進行擴充；此外更基於此平台建置最佳化功能，透過粒子最佳化演算法(Particle Swarm Optimization，簡稱PSO)作為最佳化架構，同時訓練機器學習模型(Xgboost)以擬合電磁有限元素分析(Finite Element Analysis, FEA)軟體計算結果，可快速計算出設計者所需之性能並套用於PSO架構中，不僅可快速搜尋最佳的參數，亦降低調整參數的次數以減少開發時間。

With the rapid technology advancements and the conscious awareness of global warming, electric motors have become an important device and is seen in majority of machineries including those used for everyday living. As a result, the demand for electric motors have become increasingly high, and the use of existing design and analysis processes has become challenging in meeting the timeline of demands. Generally, manufacturers usually use the existing motor specifications to adjust simple parameters through trial and error to meet the specifications and characteristic requirements of different customers. This is mainly limited by the long design time. The amount of manpower, time and energy required to design and iterate simulations using the existing method has become challenging. Also, professional know-how and experience are required to vary design parameters to yield various desired motor performances.
In order to solve the above problems, this study aimed at the needs of the existing motor design processes for Surface mounted Permanent Magnet Motors (SPMMs) to develop an automated process script platform. Meanwhile, each function is modularized to reduce the inconvenience of operation in the future. Through this modular platform, the automation script can be expanded for other different types of motors. Also, based on this platform, the optimization function can be built using the Particle Swarm Optimization algorithm PSO, as the optimization framework, while training the machine learning model (Xgboost) to fit the electromagnetic finite element analysis FEA, results. The proposed platform can quickly calculate the performance desired by the designer and apply it to the PSO. In the architecture, not only that the best parameters can be quickly searched, but also, the number of parameter adjustments can be reduced thereby reducing the development time.

[1]Grand view research, Electric Motor Sales Market Size, Share & Trends Analysis Report By Application, By Power Output (Integral HP Output, Fractional HP Output), By Motor Type (Hermetic, AC, DC), By Region, And Segment Forecasts, 2021 – 2028.
[Available]:https://www.grandviewresearch.com/industry-analysis/electric-motor-market (october 31,2022)
[2]Grand view research ,Simulation Software Market Size, Share & Trends Analysis Report By Component (Software, Service), By Deployment (On-premise, Cloud), By End Use (Healthcare, Industrial), By Application, And Segment Forecasts, 2022 – 2030.[Available]:https://www.grandviewresearch.com/industry-analysis/simulation-software-market (october 31,2022)
[3]Businesseswire, Global Simulation and Analysis Software Market Forecast to Showcase a CAGR of Over 11% Through 2021: Technavio.[Available]:https://www.businesswire.com/news/home/20161230005027/en/Global-Simulation-and-Analysis-Software-Market-Forecast-to-Showcase-a-CAGR-of-Over-11-Through-2021-Technavio(october 31,2022)
[4]C. Zhang, L. Chen, S. Yu, X. Ma, X. Wang and R. Tang, “Performance Characteristics of High Speed Permanent Magnet Machine with Different Rotor Retaining Sleeve,” 2019 22nd International Conference on Electrical Machines and Systems (ICEMS), 2019, pp. 1-5, doi: 10.1109/ICEMS.2019.8921581.
[5]E. Bradley and P. Chung, "AI tools: development trends and selection," IEE Colloquium on Strategic Research Issues in AI in Engineering, 1990, pp. 8/1-8/3.
[6]J. Song, F. Dong, J. Zhao, H. Wang, Z. He and L. Wang, "An Efficient Multiobjective Design Optimization Method for a PMSLM Based on an Extreme Learning Machine," in IEEE Transactions on Industrial Electronics, vol. 66, no. 2, pp. 1001-1011, Feb. 2019, doi: 10.1109/TIE.2018.2835413.
[7]Y. Yang, N. Bianchi, G. Bacco, S. Zhang and C. Zhang, "Methods to Reduce the Computational Burden of Robust Optimization for Permanent Magnet Motors," in IEEE Transactions on Energy Conversion, vol. 35, no. 4, pp. 2116-2128, Dec. 2020, doi: 10.1109/TEC.2020.3016067.
[8]M. Onsal, B. Cumhur, Y. Demir, E. Yolacan and M. Aydin, "Rotor Design Optimization of a New Flux-Assisted Consequent Pole Spoke-Type Permanent Magnet Torque Motor for Low-Speed Applications," in IEEE Transactions on Magnetics, vol. 54, no. 11, pp. 1-5, Nov. 2018, Art no. 8206005, doi: 10.1109/TMAG.2018.2832076.
[9]林揚笙，「應用類神經網路與最佳化演算法於馬達參數設計」，國立成功大學電機設計與驅動產業碩士班，碩士論文，2020。
[10]茆尚勳，「直驅式跑步機用直流無刷馬達之設計」，國立成功大學機械工程研究所，碩士論文，2002。
[11]王伯恭，「直驅式輪轂馬達於輕型電動機車之合設計」，國立成功大學機械工程研究所，碩士論文，2009。
[12]Hanselmen, Duane C., Brushless Permanent-Magnet Motor Design, The Writers' Collective, 2003.
[13]L. I. Jusoh and E. Sulaiman and F. S. Bahrim and R. Kumar, Design Comparison of Inner and Outer Rotor of Permanent Magnet Flux Switching Machine for Electric Bicycle Application, IOP Conference Series: Materials Science and Engineering, 2017
[14]Differences between inner and outer rotor motor | Electric motors | Engineering | Technology. [Avalible]:https://www.youtube.com/watch?v=TeJ1KkX76rg&ab_channel=murali.today(october 31,2022)
[15]余守龍，「高效率永磁無刷馬達之設計與分析」，國立成功大學系統及船舶機電工程學系，碩士論文，2009。
[16]陳育俊，「高速主軸永磁無刷馬達之設計與分析」。碩士論文，逢甲大學電機工程所，2012
[17]MPCO magnetic products Performance Curve : Example of a SmCo BH curve[Avalible]:https://mpcomagnetics.com/kownledge/performance-curve/(october 31,2022)
[18]中鋼公司，電磁鋼捲[Avalible]:https://www.csc.com.tw/csc/pd/doc/spec_es_c_2016.pdf(october 31,2022)
[19]林鳴志，模擬自動化介紹，Ansys 自動化課程
[20]PyAEDT documentation[Avalible]:https://aedt.docs.pyansys.com/(october 31,2022)
[21]黃品嘉，「應用AI於繞線機智程參數最佳化設計與實驗驗證」，國立成功大學電機設計與驅動產業碩士班，碩士論文，2021。
[22]Supervised Machine Learning: Regression and Classification, Andrew Ng [Avalible]:https://www.coursera.org/learn/machine-learning(october 31,2022)
[23]iT邦幫忙，10程式中，資料分布與離群值處理 [Avalible]:https://ithelp.ithome.com.tw/articles/10278000?sc=iThelpR(october 31,2022)
[24]CFI Education Inc. ,Skewness [Avalible]:https://corporatefinanceinstitute.com/resources/data-science/skewness/(october 31,2022)
[25]Scikit learn, MinMaxScaler [Avalible]:https://scikit-learn.org/stable/modules/generated/sklearn.preprocessing.MinMaxScaler.html(october 31,2022)
[26]T. Chen, C. Guestrin,” XGBoost: A Scalable Tree Boosting System”,in Proc. 22 nd ACM SIGKDD Int. Conf. Knowl. Discovery Data Mining,pp785-794,2016.
[27]F. Marini, B. Walczak,”Particle swarm optimization (PSO). A tutorial,Chemometrics and Intelligent Laboratory Systems”,Volume 149,Part B,2015,PP 153-165,ISSN 0169-7439.
[28]N. Yusup, A. M. Zain, S. Z. M. Hashim,Overview of PSO for Optimizing Process Parameters of Machining,Procedia Engineering,Vol. 29,2012,pp 914-923,
[29]Implementing the Particle Swarm Optimization (PSO) Algorithm in Python [Avalible]:https://medium.com/analytics-vidhya/implementing-particle-swarm-optimization-pso-algorithm-in-python-9efc2eb179a6(october 31,2022)
[30]K.T. Chau, Electric Vehicle Machines and Drives: Design, Analysis and Application, WILEY, 2015.