A Novel Adaptive Dynamic Taylor Kriging and Its Application to Optimal Design of Electromagnetic Devices
Bin Xia, Tae-Woong Lee, Kyung Kook Choi, Chang Seop Koh · IEEE Transactions on Magnetics · 2015
An adaptive dynamic Taylor Kriging (ADTK) is developed and combined with the particle swarm optimization algorithm to get a numerically efficient optimization strategy. For given sampling data, the ADTK always minimizes its fitting error without regard to a problem through optimal selection of basis functions among Taylor polynomials. An adaptive sampling method is also proposed based on the fitting error estimation, through which a minimum number of sampling data for a desired level of fitting error can be controlled. The proposed approaches are tested with an analytic function and the TEAM 25.