Design optimization of single-sided axial flux permanent magnet machines by differential evolution

Abstract

Listen

Translate article

In this paper, a design optimization approach for single-sided axial flux permanent magnet (AFPM) machines using a differential evolution algorithm for is presented. The objectives of the design optimization are to maximize the output torque per unit cost (Nm/$) and maximize the efficiency. A paremetric 2-D FEA model of an AFPM is built. A sensitivity study of design variables is carried out to determine the correlation between the design variables and the objectives, enabling the removal of insignificant design variables. Design constraints including geometrical and operating limits are considered. A total of five independent variables are employed in the optimization process. The optimization result is compared with a prototype design and results verified by 3-D FEA simulations. Index Terms—axial flux permanent magnet (AFPM) machines, design optimization, differential evolution, sensitivity study. the flux weakening area. In (10), an optimal design practice for an IPM machine with a modular stator structure based on finite element analysis (FEA) and different evolution is discussed. Both a single and a multi-objective of maximize the torque and minimum THD of back EMF process is implemented. In (11), an automated machine design process with differential evolution techniques is proposed to maximum the torque and efficiency. In (12), (13), a bi-objective optimization of a PM machine with 11 parameter variables using computationally efficient-FEA and differential evolution, to minimize torque ripple and maximize the torque per unit volume is presented. Four different machine topologies are evaluated by comparing their Pareto-optimal design sets. In (14), a multi-objective optimization of a surface PM motor with 5 variables is used to minimize of total weight and maximize a goodness function, which is defined as torque per root square of losses at rated load. The results using differential evolution(DE) is compared with results using the response surface(RS) method. It is shown that DE has a better capability for dealing with a large number of candidate designs. In (15), an optimal design by differential evolution of a surface PM machine with 8 variables and the objective of minimzing the cost of active materials per unit efficiency is presented. Stopping criteria for the DE algorithm are discussed based on both the solution space and thedesign space. In (16), a combined design of Design of Experiments and DE is implemented for the optimization of a 12 slot,8 pole, spoke type ferrite permanent magnet machines. This paper will provide insights into the design optimization for axial flux machines by means of a multi-objective differ- ential evolution algorithm. A sensitivity study of the design variables is studied. A total of five independent significant variables are employed in the design. Design objectives are to maximize the output torque cost(Nm/$) and efficiency. Optimization results are compared with a prototype design.