A design of neural-net based controllers with internal model structure for nonlinear systems

Since most process systems have nonlinearities, it is necessary to consider controller design schemes to deal with nonlinear systems. In this paper, a new neural-net based controller is proposed, which has an internal model structure. The internal model consists of the linear nominal model and the neural network. The linear nominal model and the neural network respectively work for the purpose of compensating the linear and the nonlinear components included in the controlled object. The pole-assignment control system is constructed for the augmented system which is composed of the controlled object, the internal model and the linear nominal model. Finally, the effectiveness of the newly proposed control scheme is numerically evaluated on a simulation example.