抄録
This paper proposes a design method for an input tracking control system using a neural network. A multi-layered neural network is first trained to model the input-output mapping of a controlled object of unknown, discrete-time nonlinear system. Next, this neuro emulator is used for generating Jacobian information with respect to the controlled object. A simple control law is determined by Lyapunov's method so that the output of the controlled object can follow a reference input. The advantages of this method are: 1) The neural network can emulate the output of the controlled object, and we need no mathematical model of the controlled object. 2) Training of the neural network is carried out off-line, and the control law is determined by Lyapunov's method. Thus, the system output can track a reference input in the sense of stability of linearized system. The simulation results showed the effectiveness of the proposed method. We applied also this method to an input estimating problem, as a dual problem of the tracking problem. The simulation results for this problem are also given.
