To improve the ratio between computational costs and accuracy, adaptive mesh refinement for the solution of nonlinear magnetostatic field problems is used. When the integral equation method is used to solve the field problem, multiple local refinement indicators are necessary. An expert system is necessary to combine the knowledge of these multiple indicators. Therefore, expert systems based on fuzzy logic as well as on neural networks have been developed. They are compared with respect to efficiency and accuracy with the setup of TEAM Workshop Problem #20.
展开▼