基于本构神经网络橡胶隔振器动态特性建模

Modeling of Dynamic Behaviors of Rubber Isolators Using Constitutive Model and BP Neural Network Model

提出一种基于本构和BP神经网络模型的橡胶隔振器动态特性的建模方法并验证模型的有效性。首先,分别建立本构模型、BP神经网络模型和基于本构和BP神经网络的联合模型,来表征橡胶隔振器的动态特性。第二,利用最小二乘法对本构模型进行参数识别;以实测的动刚度和滞后角为训练样本,对BP神经网络模型进行模型辨识。计算结果表明,基于本构和BP神经网络的联合模型能更好地表征橡胶隔振器动态特性的振幅相关性和频率相关性。最后,计算3种模型受到随机位移激励的时域响应,并进行试验验证。试验结果表明,基于本构和BP神经网络的联合模型能更加准确地描述橡胶隔振器的动态特性。

A modeling method for dynamic behaviors of rubber isolators using constitutive model and BP neural network model is proposed and validated.Firstly,the constitutive model,BP neural network model and the joint model consisting of both the models are established respectively to characterize the dynamic behaviors of the rubber isolators.Then,the least square method is used to recognize the parameters of the constitutive model.The BP neural network model and the joint model are trained using the measured dynamic stiffness and the loss angle.The calculation results show that the joint model can better characterize the amplitude and frequency correlation of the dynamic behaviors of rubber isolators.Finally,the time-domain responses of the three models excited by random displacement are calculated and experimentally validated.The measured results show that the joint model consisting of constitutive model and BP neural network can more accurately describe the dynamic behaviors of rubber isolators.