Ling, Poh Ping (2018) Analytical Modelling And Efficiency Optimisation Of Permanent Magnet Synchronous Machine Using Particle Swarm Optimisation. Masters thesis, Universiti Sains Malaysia.
|
PDF
Download (183kB) | Preview |
Abstract
Reka bentuk mesin biasanya merupakan proses yang rumit dengan pembolehubah yang saling berkait dan juga bergantung pada faktor-faktor lain seperti hubungan tak linear antara parameter, sifat bahan, batas rekabentuk dan keperluan aplikasi. Pemodelan beranalisis masih diperbaiki secara berterusan untuk menghasilkan ramalan yang menyerupai analisis unsur terhingga (FEA) dan operasi mesin masa-nyata. Akan tetapi, kedua-dua pemodelan beranalisis serta FEA tidak dapat mengenal pasti parameter mesin diperlukan untuk pengoptimuman kecekapan mesin yang berbeza. Pemilihan pembolehubah secara stokastik juga tidak cekap dalam pengoptimuman mesin kerana hubungan-hubungan tak linear antara pembolehubah mesin PMSM. Lantaran itu, penyelidikan ini tertumpu kepada penggunaan pemodelan subdomain beranalisis medan magnetik dan sifat-sifat berkaitan untuk pengoptimuman tiga-fasa, 12-lubang alur/8-kutub PMSM lekap permukaan dengan topologi pemutar luaran. Teknik ini telah digunakan dengan mengubah nilai-nilai pembolehubah mesin terpilih seperti lengkuk kutub magnet, ketebalan magnet, kelebaran sela udara, dan bukaan lubang alur. Selepas itu, suatu algoritma berkomputer pintar Pengoptimum Kerumunan Zarah (PSO) digunakan untuk mengubah pembolehubah mesin yang terpilih secara serentak, bagi mencari penyelesaian optimum berkrompomi untuk prestasi mesin yang tertinggi. Hasil prestasi mesin yang terbaik adalah berdasarkan indeks prestasi yang terpilih – kecekapan dan THDv. Hasil yang didapati daripada pemodelan beranalisis dan Pengoptimum Kerumunan Zarah (PSO) telah dsahkan dan dipersetujui dengan FEA. Daripada hasil penyelesaian PSO, keempat-empat pembolehubah reka bentuk mesin yang dioptimumkan berjaya mengoptimumkan prestasi mesin. Hasil penyelidikan ini menunjukkan gabungan pemodelan beranalisis dan PSO dapat memudahkan kaedah jangkaan stokastik dalam pembolehubah mesin serta menjadikan proses reka bentuk dan pengoptimuman reka bentuk mesin yang lebih cekap. _______________________________________________________________________________________________________ Machine design has always been a comprehensive process with inter-dependant variables that are subjected to many factors such as non-linear relationship between parameters, material properties, design limitations and application-dependant requirements. While analytical modelling has been continuously developed to predict as closely as possible to resemble the finite element analysis (FEA) and real-time machine operation, but analytical modelling as well as FEA are unable to pin-point specific machine variables required to be optimised for a particular design. Furthermore, stochastically choosing machine variables is not efficient in machine optimisation as there are complicated non-linear relationships between machine parameters in PMSM. Therefore, this research focuses on the usage of subdomain modelling for analytical prediction of magnetic field and other attributes to optimise a three-phase, 12slot/8pole surface mounted PMSM with external rotor topology, by varying selected machine variables - magnet pole arc, magnet thickness, air-gap width and slot opening individually. Subsequently, an intelligent computational algorithm - Particle Swarm Optimization (PSO) was later applied to all the machine variables simultaneously to find the optimal solution for a compromised optimal machine performance. The improved machine performace are based on the chosen performance indexes – efficiency and THDv. The results obtained from the analytical prediction and particle swarm PSO were compared with FEA for verification and was found to be in good agreement. From PSO study, the four machine design variables has been simultaneously optimised and successfully produced parameters for a performance-optimised machine. The research results has also demonstrated that by simplifying traditional stochastic methods in the targeted machine variables, a combination of analytical modelling and PSO allows a more efficient machine design and optimisation process.
| Item Type: | Thesis (Masters) |
|---|---|
| Uncontrolled Keywords: | Full text is available at http://irplus.eng.usm.my:8080/ir_plus/institutionalPublicationPublicView.action?institutionalItemId=4642 |
| Subjects: | T Technology T Technology > TK Electrical Engineering. Electronics. Nuclear Engineering > TK1001-1841 Production of electric energy or power. Powerplants. Central stations |
| Divisions: | Kampus Kejuruteraan (Engineering Campus) > Pusat Pengajian Kejuruteraaan Elektrik & Elektronik (School of Electrical & Electronic Engineering) > Thesis |
| Depositing User: | Mr Mohd Jasnizam Mohd Salleh |
| Date Deposited: | 13 May 2019 02:35 |
| Last Modified: | 13 May 2019 02:35 |
| URI: | http://eprints.usm.my/id/eprint/44311 |
Actions (login required)
 |
View Item |
