Modularly structured B-spline networks for internal model control

A method is introduced to weaken the problems arising when a complex nonlinear process has to be controlled. The proposed method is based on the use of modularly structured B-spline networks (MBSN) which are integrated in an internal model control (IMC) structure. It is shown that the networks are suitable for this control strategy due to their simple mathematical description allowing model inversion in an easy way. An example is given to illustrate the advantages over other neural based methods using separate networks considered as controllers which have to be trained additionally to the process modelling network