Receding horizon implementation of optimal servo problem: application to a nonlinear process with input multiplicities

A receding horizon control technique is developed for nonlinear systems which exhibit input multiplicity behaviour. Based on local linearization of the nonlinear process model around its trajectory, an infinite-time horizon linear quadratic optimal servo problem is formulated with penalty on the control input as well as on its time-derivative in the performance index, so that it can handle a singular static gain situation. The local linearization approach is shown to be asymptotically stable and achieve a zero steady state error response at the operating point with a nonzero static gain. At the zero static gain operating point, it is shown that the closed loop can be made asymptotically stable by giving an “unreachable” set-point