NON-LINEAR CONTROL USING PARAMETER ESTIMATION FROM FORWARD NEURAL MODEL

Anna Jadlovská

   A neural network application for system identification and control of non-linear process is described in this paper. The non-linear identification is mostly using feed-forward neural networks as a useful mathematical tool to build a non-parametric model between the input and output of a real non-linear process. The possibility of on-line estimation of the actual parameters from an off-line trained neural model of the non-linear process using the gain matrix is considered in this paper. This linearization technique is used in the algorithm of on-line tuning of the controller parameters based on a pole placement control design for the non-linear SISO process.

Keywords: neural model, NARMAX structure, non-linear parameter estimation

[full-paper]