Optimal Design of Novel Pole Piece for Power Density Improvement of Magnetic Gear Using Polynomial Regression Analysis

A magnetic gear that is capable of noncontact torque transmission has replaced mechanical gear, and has advantages of high efficiency and improved reliability. However, it also has the problem of a low torque density, which requires improvement. In this paper, a novel pole piece shape is proposed in order to improve the problem of low torque density of the magnetic gear. In order to identify the functional relationship between design variables of the proposed pole piece shape and response variables such as torque and torque ripple, response surface methodology is used, and the Box-Behnken design is used to establish the design of experiment. In addition, using variance analysis and regression analysis of design variables and response variables, response surface of the response variable to the design variable, and its response surface equation are estimated to derive optimal design variables for the proposed pole piece shape. The experiment data required for predicting the relationships among them are obtained using finite-element method based on two-dimensional (2-D) numerical analysis. Therefore, this paper derived an optimal model for the magnetic gear with the novel pole piece using the Box-Behnken design, and the validity of the optimal design of the proposed pole piece shape through variance analysis and regression analysis demonstrated.