Decentralized adaptive robust controller design for complex system based on partition of unity

Abstract

A new method for designing decentralized adaptive robust controllers was proposed which focuses on a class of more general uncertain complex systems, using the concept of the partition of unity in differential geometry to deal with system uncertainties. In this method the uncertainty of the system to be controlled was normalized firstly, and then the partition of unity that was subordinated to an open covering of state variables compact set was constructed. Subsequently the approximation was realized by using its property that can approximate nonlinear continuous function with arbitrary precision, and then the decentralized adaptive robust controller of complex systems and adaptive laws of approximate parameter estimation were designed. Compared to existing methods, the proposed algorithm requires simpler assumed conditions and no complicated computations. Simulation result shows that the method is valid.