In this paper, a self-constructing fuzzy wavelet
neural network (SFWNN) is used to approximate an unknown
nonlinear term in the system dynamics with the structure and
parameter learning abilities concurrently. Further, a selfconstructing
fuzzy wavelet neural control (SFWNC) system,
which is composed of a computation controller and a robust
compensator, is proposed. The computation controller using the
SFWNN approximator is the main controller and the robust
compensator is designed to eliminate the effect of the
approximation error. All controller parameters of the SFWNC
system are adaptively update based on the Lyapunov stability
theory and the projection algorithm to guarantee the closed-loop
system stability. Finally, the effectiveness of the proposed
SFWNC system is verified by simulation results and the SFWNN
approximator has the admirable property of small fuzzy rules
size and high learning accuracy.
Relation:
Joint 17th World Congress of International Fuzzy Systems Association and 9th International Conference on Soft Computing and Intelligent Systems