An efficient and robust method for parameterized nonintrusive reduced-order modeling

摘要

A method of constructing parameterized nonintrusive reduced-order models (NIROMs) is given. The approach relies on a geometrical interpretation of NIROM, requires only a single layer of interpolation to be applied for both system state and parametric dependence of the model and is applicable to systems characterized by any number of parameters spanning arbitrary orders of magnitude. The method is applied to three representative test problems and evaluated in terms of accuracy and speed, showing good performance.