Optimization of Permanent Magnet Assemblies Using Genetic Algorithms

This paper describes the utilization of genetic algorithms to optimize the topology of assemblies of permanent magnets. Conventional genetic algorithms are modified to employ pixel refinement to improve the computational efficiency of the optimization process. An example case is demonstrated for optimizing the figure of merit of a permanent magnet dipole. By using only four magnetization directions, a figure of merit value of 0.11 is achieved using the proposed optimization method.