Robust Design Optimization of Permanent Magnet Synchronous Machine Utilizing Genetic and Taguchi's Algorithm

Optimization of permanent magnet synchronous machine (PMSM) in high performance applications is a multidimensional problem. Smooth torque production is an important performance objective for low torque ripple applications. Moreover, reduction in part-to-part variations and reduced sensitivity towards manufacturing tolerances and variations are important factors to have high yield in mass production. This paper utilizes optimum design methodologies along with Taguchi's robust design techniques to develop an optimum and robust motor design. Genetic algorithms are used to optimize the selected design parameters due to their faster searching capability and then Taguchi's robust design techniques are used to select the appropriate design parameter values for mass production. The paper considers practical noise conditions such as eccentricity, magnet placement and magnet property variations to apply the dynamic robust design techniques where current is considered as the input. Finite element model is developed to run the design of experiments.