Permanent magnet motor design optimisation for sensorless control

This paper looks at a novel optimisation approach to the design of surface mounted permanent magnet (SMPM) machines with self-sensing capabilities. A methodology will be presented which will look at the use of genetic algorithms (GA) to contemporarily maximise the output torque and the self sensing properties of such machines. A GA optimisation environment has been grafted with a finite element analysis (FEA) environment to enable the designer to account for both geometrical and saturation saliencies for an effective determination of the machine's self sensing characteristics. Satisfactory results were obtained in terms of torque maximization and self sensing capability. Sensitivity of the major geometrical parameters of the machine will be discussed in terms torque density and the self-sensing and sensorless design and performance for high loading conditions will be considered.