Analysis of Multi-objective Optimized Design Method of Permanent Magnet Synchronous Motor

• Main Authors: Shao-Huai Jiang, 蔣少淮
• Other Authors: Kuo-Hua Huang
• Format: Others
• Language: zh-TW
• Published: 2013
• Online Access: http://ndltd.ncl.edu.tw/handle/22420758349091959630

碩士 === 國立勤益科技大學 === 電機工程系 === 101 === This thesis aims to multi-objective optimized design of　permanentmagnet synchronous motor use a 3D finite element　analysis of　electromagnetic field software Ansoft Maxwell 3D　to reduce the cogging　torque, torque ripple and enhance the　operational efficiency. In which the　Taguchi's Orthogonal　Array and the Response Surface Design's Central Composite　Design are used in experimental designs, and then sequentially　with Regression Analysis and Artificial Neural Network for　building　predictive models to find the optimal structure. First, in order to facilitate experimental designs by Orthogonal Array and Central Composite Design, according to the prototype electromagnetic simulation results choose various design parameters and criteria from influence design objectives on motor. Second, using Regression Analysis and Artificial Neural Network design of experimental software Design Expert and CAFE build predictive models, respectively. Finally, a 3D finite element analysis of electromagnetic field software Ansoft Maxwell 3D, is conducted to verify and comparison results of optimal structure. The result indicated that all the single objective optimal structure and the multi-objective optimal structure by Taguchi-Analysis of Variance, Taguchi Regression Analysis, Taguchi-Artificial Neural Network, Response Surface-Regression Analysis and Response Surface Artificial Neural Network are the same. Therefore, when design parameters for design objectives are monotonous situations, any optimization methods have accurate prediction effect to the optimal structure.