Spline based particle swarm optimization of a synchronous reluctance machine rotor

Increasing access to advanced computing techniques is opening opportunities to incorporate automated design methods to the optimization of electrical machines. Common to many topology optimisation approaches is a long simulation time and to address this, conventional methods constrict the design domain. This paper considers a particle swarm optimization with a novel Bezier curve approach to explore the design space without priori assumption, reducing the degrees of freedom of the problem without restricting the design domain providing opportunities for further improvement. Results show successful validation of the technique by comparison to an analytical solution and convergence towards a synchronous reluctance machine when torque and air-gap flux densities are considered in the cost function. The methodology is demonstrated to be able to tackle the multi-objective non-linear problem of the reluctance rotor, in successfully generating low torque ripple designs. Further work is proposed to implement multiple internal features and advanced computing methods to better explore the design space in a reduced time with the aim of generating designs that can compete with conventional methods.