STATE ESTIMATION AND CONTROL OF NONLINEAR PROCESS USING NEURAL NETWORKS

Anna Jadlovská

   This paper considers the use of neural networks for non-linear state estimation, identification and control of non-linear processes. The non-linear identification is using feed-forward neural networks as useful mathematical tool to build model between the input and the output of a non-linear process. In this paper is considered the possibility an on-line state estimation of the actual parameters from off-line trained neural state space model of the non-linear process using the gain matrix. This linearization technique is used in the algorithm on-line tuning of the controller parameters based on pole placement control design for non-linear process.

Keywords: dynamic neural models, non-linear state estimation, gain matrix, non-linear control

[full-paper]