MAGNET DESIGN OPTIMIZATION USING VARIABLE METRICS

The optimal design of a magnet assembly is approached using a three step process. First, the key parameters within the objective performance index are computed For the variation range of the problem. Second, the performance index is fitted to a smooth polynomial involving products of the powers of all variables. Third, a constrained optimization algorithm is employed to predict the optimal choice of For the variables. An assessment of the integrity of the optimization program is obtained by comparing the final optimized solution with that predicted by the field analysis in the final configuration Additional field analysis is recommended around the final solution to fine tune the solution.