Nonlinear Structural Vibration Suppression Using Dynamic Neural Network Observer and Adaptive Fuzzy Sliding Mode Control

Abstract

Nonlinear vibration control of structures has attracted much attention in recent years. In this paper, the adaptive fuzzy sliding mode (AFSM) control algorithm is adopted to actively control nonlinear structural vibration. Since the AFSM control algorithm needs the full state feedback of the structure, a dynamic neural network (DNN) observer is proposed, considering the nonlinearity of the structure. The neural network weights are adapted on-line, and no off-line learning is required. Furthermore, no exact knowledge of structural nonlinearities is needed. The weight training algorithm is established based on Lyapunov stability theory in the presence of modeling error. A semi-active control algorithm is used for MR damper to track the active control force, which is computed by the AFSM control algorithm based on the DNN observer. Results of the numerical simulation of semi-active control of a 20-story nonlinear building using MR dampers verify the performance of the DNN observer and the effectiveness of the proposed AFSM control.