Short horizon optimal control of nonlinear systems

An optimal model predictive controller for nonlinear discrete-time systems is considered. A block input-state realization of the system, which transforms it to a square system with equal number of inputs and states, is used to develop a receding horizon optimal controller designed to decrease computational expense in relation to more traditional optimal controllers. A nonlinear feedback control law is designed where a neural network in the feedback loop is used to generate an optimal control input. The generated input approximates a solution that is minimal with respect to a quadratic cost function with parameters governing the final states and the magnitude and variation of the control input, for either regulation or tracking. A local stability and robustness analysis of the controller is also presented. A simulation example shows use of the control method.