Periodic disturbance rejection of nonlinear systems via output feedback with Neural Network approximation

In this paper neural network (NN) is applied for rejecting periodic disturbances in output feedback nonlinear system. The NN adopted here is Adaptive Radial Basis Function Neural Network (ARBFNN). The parameters of the system, except the high gain frequency, and disturbance are assumed to be unknown. We also postulate that the uncertainty of the output feedback system is bounded by an existing unknown constant polynomial and then adaptive technique can be employed. All of the unknown parameters in the system are dealt with by adaptive control techniques. Control design via backstepping approach is used for this high order system case. The uniform stability is guaranteed through Lyapunov analysis and the tracking error is restricted to an acceptable small region around the origin. An example is included to demonstrate the feasibility of the proposed theory.