A robust adaptive cancellation scheme for nonlinear uncertain system based on extended plant model

An extended control plant is introduced to substitute the plant used in ordinal adaptive cancellation system. All of the inputs and outputs of such an extended plant are online available and hence its model can be identified online conveniently. Based on the static nonlinear feedforward neural network model of this extended plant, a new adaptive cancellation system structure and algorithm are proposed. In this system, a dynamic nonlinear feedforward neural network is used for controller. The control error is back propagated through the extended plant model to update the connection weights of controller. It is shown by simulation results that the proposed adaptive cancellation scheme is effective for both linear and nonlinear system. Moreover, since both the plant model and controller are online adjusted simultaneously, this control scheme has good robustness for the variation of system parameters and features.