Design Of Neurocontrollers Based On Approximate Optimum Control For Restricted Processes

In this paper, the computing procedure of controllers for processes modelled as nonlinear dynamical systems is detailed. The technique is based on optimal control, which allows to work with arbitrary constraints and non-quadratic cost-to-go functional. Simple conditions are required to the performance index, such as convexity and continuity. The proposed technique to solve the optimal control problem for nonlinear processes with constraints is known as approximate dynamic programming, and here are given details of its implementation in processes with different characteristics, where the change in the complexity is observed as that system requires larger dimensions to be represented.