PREDICTIVE VARIABLE STRUCTURE CONTROL FOR A CLASS OF NONLINEAR SYSTEM

Nonlinear sliding mode predictive controller is designed for a class of nonlinear system with unmodeled dynamic characteristics and nonlinear term. The method is based on nonlinear optimal predictive control. The variable structure control law minimizes the quadratic index of a predictive sliding mode, which contains the cost function of control preventing the control effect from saturation for in most practical implementation the control inputs are bounded by physical constraints and energy constraints. According to the immeasurable states, the variable structure observer for nonlinear systems is adapted. The variable structure system method is apt to the realization of observer with variable parameters and uncertainty. The proof shows that the states of the observer asymptotically convergence to the real states of the system although it is of uncertainty and nonlinear terms. Finally, the digital simulation results prove the effectiveness of the proposed method.

PREDICTIVE VARIABLE STRUCTURE CONTROL FOR A CLASS OF NONLINEAR SYSTEM

Nonlinear sliding m ode predictive controller is designed for a class of nonlinear system w ith unm odeled dynam ic characteristics and nonlinear term . The m ethod is based on nonlinear opti- m alpredictive control. The variable structure controllaw m inim izes the quadratic index ofa predic- tive sliding m ode, w hich contains thecostfunction ofcontrolpreventing the controleffectfrom satu- ration for in m ostpracticalim plem entation the controlinputs are bounded by physicalconstraints and energy constraints. According to the im m easurable states, the variable structure observer for nonlin- ear system sisadapted. The variablestructure system m ethod isaptto therealization ofobserverw ith variable param eters and uncertainty. The proofshow s thatthe states ofthe observer asym ptotically convergence to the realstates ofthe system although itisofuncertainty and nonlinear term s. Final- ly, the digitalsim ulation results prove the effectiveness ofthe proposed m ethod.