Neural network based dynamic pole placement control of nonlinear systems

A novel algorithm for control of nonlinear discrete time systems is presented in the paper. The algorithm is based on dynamic feedback linearization of controlled nonlinear system. Linearized closed loop system is equivalent to a predefined discrete time transfer function representing reference model for the control system. Parameters of the transfer function are defined as nonlinear functions of the current control error and the derivative of the control error. Thus poles of the closed loop system are placed dynamically according to the predefined surface of poles providing necessary dynamics of the control system. Linearizable model of the controlled nonlinear system can be identified by training an artificial neural network of the specific structure. This model is applied to dynamic pole placement control of the identified nonlinear system. The proposed control technique can be applied to a wide class of nonlinear systems. The effectiveness of the proposed control technique is demonstrated on numerical example.