DYNAMICAL NEURAL NETWORK OBSERVER DESIGN FOR THE NONLINEAR VIBRATION MODELS OF STRUCTURES

摘要

Due to the complexity of nonlinear system, the control of nonlinear vibration of structure always needs the total state feedback, but it is difficult to get all the states of structure. In this paper, the intelligent observer for the nonlinear vibration of structure using dynamical neural network is proposed to get the states needed to implement the structural control. The neural network weights are tuned on-line, with no off-line learning required. The robustness of algorithms for network weights adjusting based on Lyapunov theory is established in the presence of a modeling error term and a robust control term. The uniform ultimate boundedness of state observation errors is guaranteed. Simulation results made on the twenty- story steel frame model which is referred to as the third generation Benchmark model are given to demonstrate the effectiveness of the proposed dynamical neural network observer algorithm.