Application of RBFEM in the fast design optimization electric field

Some methods that combine FEM (finite element method) and nonlinear programming are applied to optimization design of high voltage electric field. The use of FEM as a cost function evaluator is constrained by the sequential processes of geometry construction, meshing, and solving in the post-processing stages. The computational overhead makes FEM-based optimization unattractive despite its accuracy. So the RBFEM (Reduced-basis FEM) is introduced. The principle of RBFEM is analyzed and the program diagram is also proposed. The order of reduced-basis matrix is much smaller than the number of nodes, and its sets of data values cover the entire regime including all FEM solutions. Owe to compute the reduced-basis matrix the computational effort can be cut down without significantly affecting the accuracy of the FEM solution. The novel algorithm is applied to design the high voltage winding inside 100 kV transformer with SF₆ insulation. The ratio of the maxim and minimum of electric field intensity is reduced from 1.79 to 1.15.