Adaptive observer backstepping control using neural networks

This paper extends the application of neuro-control approaches to a new cla ss of nonlinear systems diffeomorphic to output feedback nonlinear systems with unmeasured states. A neural-based adaptive observer is introduced for state estimation as well as system identification using only output measure ments during on-line operation. System identification is achieved via the o n-line approximation of a priori unknown functions. The controller is desig ned using the backstepping control design procedure. Leakage terms in the a daptive laws and nonlinear damping terms in the backstepping controller are introduced to prevent instability from arising due to the inherent approxi mation error. A primary benefit of the on-line function approximation is th e reduction of approximation errors, which allows reduction of both the obs erver and controller gains. A semiglobal stability analysis for the propose d approach is provided and the feasibility is investigated by an illustrati ve simulation example.