Design optimization of a superferric octupole using various evolutionary and deterministic techniques

The paper describes different strategies for the design optimization of electromagnets. In order to compare the efficiency of these procedures, the same objective weighting function was minimized using a deterministic search algorithm, evolution strategy, and genetic algorithms. The objective function is based on minimizing the relative normal multipoles produced by the harmonic analysis of the flux density produced by the pole. The pole shape was treated differently in applying the various techniques to the problem. The optimization techniques were independently applied to optimize the magnet's shape. Results of the numerical solution as well as the harmonic analysis were compared. The methods are tested on an example of a superferric octupole magnet. Good results from the various techniques was achieved. A discussion on the practical aspects of these optimization methods is presented.