Process modelling method for neural network

摘要

The method includes the step of representing a time dependence of a first process size (PN) through a parameter-dependent differential equation with a known solution form, whereby at least the solution parameter does not depend linearly on at least one second process size. The or each non-linear function which represents the dependence of the or each solution parameter on the second process size, is modeled by device of a neural network (NN1, NN2) which was trained with measurable process sizes.