A hybrid neural network model based modelling methodology for the rubber bushing
Liangcheng Dai, Maoru Chi, Chuanbo Xu, Hongxing Gao, Jianfeng Sun, Xingwen Wu, Shulin Liang
Abstract
To fully consider the influences of the exciting frequency (0–20 Hz) and the environmental temperature (−50–20°C) on the dynamic characteristics of the rubber bushing in the railway vehicle, a hybrid neural network model is applied to develop the detailed model of the rubber bushing. The nonlinearities of the rubber bushing existing in the elastic element, the friction element and the viscous element are involved in the model to reflect the frequency- and amplitude-dependent characteristics of the rubber bushing. The experimental data are collected and exploited to find out the relationship between the input and output with the help of the neural network model. Furthermore, the influences of the environmental temperature on the dynamic parameters of the rubber bushing are considered. By testing the dynamic performance of the rubber bushing under the excitation with different frequencies, amplitudes and temperatures, the hybrid neural network model is trained and its critical parameters are identified. In this way, the trained hybrid neural network model for the rubber bushing can accurately reflect the dynamic performance in the presence of various excitation and environment temperatures.