Neural control of MIMO nonlinear plants: a gain scheduling approach

A neural controller design technique for MIMO nonlinear plants is presented. In a first stage, the plant is identified by a neural network model. In a following stage, using the neural network model, linearized models of the plant are obtained about several operating points. With the aid of conventional linear control theory, linear controllers are designed. Finally, a neural-controller is trained to replace the linear controllers. Simulation results on nonlinear plant are presented showing the applicability of the described technique.