چكيده به لاتين
Abstract:
This study investigates the tracking problem of a class of switched nonlinear delayed systems with nonstrict-feedback form. To aim this, an adaptive neural network based backstepping approach in the presence of dynamic surface control is suggested. Dynamic surface control strategy overcomes the explosion of complexity problem in backstepping approach. Due to unstructured uncertainties and unmodeled terms in many practical systems, without loss of generality, it is supposed that all nonlinear functions are unknown. A neural network approximator based on radial basis functions is utilized. It’s a common proposition that all system states are not accessible which make it is necessary to design an appropriate observer. A switched neural network based observer is proposed to observe system states. To guarantee the switched observer dynamics, a condition based bilinear matrix inequality is presented. To retain the tracking error within a predefined bound, prescribed performance bound control is utilized. Finally, according to Lyapunov-Krasovskii and average dwell time methed, the boundedness of the closed-loop signals is proved.Numerical and practical examples are used to illustrate the effectiveness of proposed approach.
Keywords: Switched nonlinear delayed systems, Nonstrict-feedback form, Backstepping, Dynamic surface control, Prescribed performance bound, Average dwell time.
