BP神经网络预测橡胶隔震支座竖向刚度  

Prediction of vertical stiffness of rubber isolation bearings by BP neural network

作　　者：陈凌阳 崔郎郎 陈永辉 王宇航[3] 柯珂 CHEN Lingyang;CUI Langlang;CHEN Yonghui;WANG Yuhang;KE Ke(College of Civil Engineering,Hunan University,Changsha 410082,Hunan,China;Citic Heavy Industries Co.,Ltd.,Luoyang 471039,Henan,China;School of Civil Engineering,Chongqing University,Chongqing 400045,China)

机构地区：[1]湖南大学土木工程学院,湖南长沙410082 [2]中信重工机械股份有限公司,河南洛阳471039 [3]重庆大学土木工程学院,重庆400045

出　　处：《建筑科学与工程学报》2023年第6期83-90,共8页Journal of Architecture and Civil Engineering

基　　金：工信部高技术船舶科研项目(MC-202014-S01)。

摘　　要：针对目前橡胶隔震支座竖向刚度理论公式不能准确计算竖向刚度的问题,建立了经试验验证的精细化ABAQUS有限元模型;在此基础上,选取常用几何尺寸(内、外径)、单层橡胶厚度、橡胶层数构造参数矩阵,基于大量有限元分析数据得出了橡胶隔震支座竖向刚度的BP神经网络预测模型;最后,基于有限元分析结果对BP神经网络模型、中国橡胶隔震支座规范和文献提出的竖向刚度计算公式进行了评估。结果表明:建立的基于内径、外径、单层橡胶厚度和橡胶层数的橡胶隔震支座竖向刚度BP神经网络预测模型精度较高;BP神经网络预测结果与试验结果的相关系数趋近于1,基于BP神经网络对橡胶隔震支座竖向刚度进行计算和预估完全可行;BP神经网络模型相对于传统拟合方法能更好解决多变量线性耦合关系。Aiming at the problem that the vertical stiffness of rubber isolation bearings can not be accurately calculated by the theoretical formula at present,a refined ABAQUS finite element model verified by tests was established.On the basis,the common geometric dimensions(inner and outer diameters),the thickness of singlelayer rubber and the number of rubber layers were selected to construct parameter matrices,and the BP neural network prediction model of vertical stiffness of rubber isolation bearings was obtained based on a large quantity of finite element analysis data.Finally,based on the finite element analysis results,the BP neural network model,China s rubber isolation bearing specifications and the vertical stiffness calculation formula proposed in literature were evaluated.The results show that the established BP neural network prediction model of vertical stiffness of rubber isolation bearings based on inner diameter,outer diameter,thickness of single rubber layer and number of rubber layers has high accuracy.The correlation coefficient between the predicted results of the BP neural network and the experimental results approaches 1,and it is completely feasible to calculate and estimate the vertical stiffness of rubber isolation bearings based on the BP neural network.The BP neural network model can better solve multivariable linear coupling relationships compared to traditional fitting methods.

关 键 词：橡胶隔震支座 竖向刚度 参数化建模 BP神经网络 

分 类 号：TU352.12[建筑科学—结构工程]