Dual Heuristic dynamic Programming for nonlinear discrete-time uncertain systems with state delay

The paper proposes a novel iterative control scheme based on neural networks for optimally controlling a large class of nonlinear discrete-time systems affected by an unknown time variant delay and system uncertainties. An iterative Dual Heuristic dynamic Programming (DHP) algorithm has been envisaged to design the controller which is proven to converge to the optimal one. The key elements required by the DHP, namely the performance index function, the optimal control policy and the nonlinear delay-affected discrete-time system are modeled with feedforward neural networks. Examples demonstrate the validity of the proposed optimal control approach and its effectiveness in dealing with nonlinear time delay situations.