Synthesis, inverse problems and optimization in computational electromagnetics

Abstract

The paper gives an overview on mathematical optimization techniques specially suited to problems in electromagnetism. It is described how efficient optimization procedures consist of decision making, treatment of non-linear constraints, and an algorithm for minimizing the objective function. The optimization of a rectangular waveguide branching, a permanent magnet synchronous machine, and the iron cross-section of a superconducting magnet are given as examples. Optimization routines are used for the tracing of manufacturing errors in the production of superconducting coils. Finally the application of a genetic algorithm to a dipole coil design is discussed as an example for field synthesis.