Frequency dependence prediction and parameter identification of rubber bushing

Abstract

Affected by frequency, amplitude and some other factors, the dynamic mechanical properties of rubber bushing are nonlinear. In order to study the frequency dependence of the rubber bushing, a BP neural network optimized by genetic algorithm (GA-BP neural network) is applied to predict the dynamic stiffness and loss factor under frequency of 61–100 Hz. The training data refers to the test data under frequency of 1–60 Hz. And the algorithm is demonstrated by the elastomer test of rubber bushing under amplitudes 0.2 mm, 0.4 mm and 0.6 mm. The results show that the prediction error of dynamic stiffness is less than 1%, and the prediction error of loss factor is less than 3%. In order to apply the predicted results to the software for simulation, a five-parameter mathematical model (FPM) consisting of three elastic elements and two damping elements is developed, and the model parameters are identified by least squares method. According to the fitting results and test data, the fitting error of dynamic stiffness is less than 2%, and the fitting error of loss factor is less than 3%. The GA-BP neural network and FPM model predict the dynamic mechanical behaviour of rubber bushing without the performance of iterative experiments and the incurrence of a high computational cost, making it applicable to analyze full-size vehicles with numerous rubber bushings under various vibration load conditions.