PROCESS IDENTIFICATION USING HAMMERSTEIN SYSTEMS

摘要

This paper considers identification of industrial processes using a Hammerstein structure. The dynamics are represented by a transfer function with unit steady-state gain. An arbitrary parameterised static mapping can represent the steady-state model for the process. Simultaneous estimation of the parameters in both dynamic and static parts is considered; the gradients required in the estimation of the parameters of the nonlinear part are given. An identification example using a binary distillation column model is given, where the static part is modelled using a 0-order Sugeno fuzzy structure. Simulations show good results.