Identification of nonlinear hysteretic systems by artificial neural network

An identification method is developed for nonlinear hysteretic systems by use of artificial neural network in the paper. Employing the Bouc–Wen differential model widely used for memory-type nonlinear hysteretic systems, the approach sets up a Bouc–Wen model-based neural network. The weights of the designed specifically network correspond to the Bouc–Wen model parameters and are thus physical ones. Taking advantage of powerful function approximation capability of neural network, the nonlinear hysteretic systems can be identified with the proposed approach by network training. The identification scheme is validated by a simulated case and thereafter applied to modeling of a wire cable vibration isolation experimental system. The results show that the presented identification method can identify the nonlinear hysteretic systems with high accuracy.

► An identification method is developed for nonlinear hysteretic systems by use of artificial neural network.
► Through building Bouc–Wen model-based neural network, all Bouc–Wen model parameters can be identified in one stage.
► The identified results of wire cable vibration isolation system demonstrate high accuracy of the present approach.