Fuzzy neural network control and identification for uncertain nonlinear systems

In this paper, the problem of identification and control of uncertain nonlinear systems is investigated based on fuzzy neural network. The considered systems are unknown and with external disturbances, so fuzzy neural networks are employed to approximate the unknown system functions. By doing this, an identification model of the controlled system can be obtained. Based on this model, a controller with adaptive mechanism can be designed for the system. The controller can attenuate the external disturbance to a given level, and guarantee the stability of the closed-loop system. Satisfactory identification and control of the system can be realized at the same time. Simulation example is given to demonstrate the effectiveness of the proposed controller.

1 INTRODUCTION

Fuzzy control is an intelligent control method as it can incorporate a priori knowledge into controller and does not need mathematics model of the controlled system for control design [1], [2], [3], [4]. Neural network is another intelligent control method, it can be trained to get desired con-trol objective. However, pure fuzzy control or neural network control cannot guarantee the performance of the controlled systems, not even the basic stability. Because of this, the development of this control method is restrictive. It is well-known that adaptive systems have the ability of online learning, it can achieve good control effect when there are unknown parameters in the control system [5], [6], [7], [8]. But it requires mathematics model of the controlled system, even though the model can contain some unknown parameters, and the unknown parameters must be linear ones. So any of these methods exists some drawbacks, though all of them have achieved many successes. The combination of them may take the advantages and compensate for the disadvantages of each method.