Joint interface stick-slip friction model parameters identification

Joint friction plays a significant role in the overall structural dynamical response. Recently multi-scale stick-slip friction models have been presented to improve dynamics prediction performance of engineering structures. In this paper, a method is proposed to identify the nonlinear joint model parameters using a back-propagation neural network. The joint in structures is represented by an Iwan model which considering multi-scale stick-slip friction behaviors. The amplitude envelope of structural time domain dynamical response is used as input parameters to the neural network for identification. The proposed method is tested on a numerical simulated nonlinear vibration model, both in the absence and presence of noise, and a lap-jointed experimental beam. It is found that nonlinear model parameters can be reasonably estimated with well accuracy by the proposed method.