Identification and Control for Nonlinear Discrete Time Systems using a New Neural Network

Y. Yamamoto (Japan)


Identification, interconnected neural network, nonlinear discrete time system, and simple model matching control.


A new identification method using an interconnected neural network is proposed in this paper for a nonlinear discrete time system with/without time lag. It is most important that the coefficient of linear input term is numerically estimated in addition to other linear and nonlinear terms. Then, the identification result is used for control to examine the effectiveness of the proposed method. The control used here is the simple model matching (SMM) method, which only requires a linear input term included. So, the proposed identification method matches with the SMM control. These give a unified approach for nonlinear discrete time systems.

Important Links:

Go Back