考虑非线性时变相关特性的支座纵向位移精细模拟方法

A Fine Simulation of Longitudinal Displacement of Bearings Considering Nonlinear Time-Varying Correlation

为准确模拟支座纵向位移、实现支座纵向伸缩性能的精准评估,以铜陵长江大桥为背景进行研究。基于该桥主梁温度和支座纵向位移的健康监测数据,分析主梁温度对支座纵向位移的非线性时变影响规律,提出基于非线性时变系数回归模型的支座纵向位移精细模拟方法;利用该方法计算支座纵向位移,并与实测数据及多元线性回归模型的模拟值进行对比。结果表明:在1 d内支座纵向位移与主梁温度之间具有较明显的非线性时变相关特性,相关特性曲线呈现出扁圆形的时滞特征;非线性时变系数回归模型的模拟结果与实测结果的变化趋势更为接近,其模拟结果与实测结果之间的绝对误差为10.5%,多元线性回归模型的模拟结果与实测结果之间的绝对误差为30.79%,非线性时变系数回归模型比多元线性回归模型具有更高的拟合精度。

The Tongling Changjiang River Bridge is used as a case to study how to accurately simulate the longitudinal displacement of bearings and realize precise assessment of longitudinal expansion and contraction performance of bearings.The data of main girder temperature and longitudinal displacement of bearings gained from the health monitoring system were adopted to analyze the non-linear time-varying correlation between the main girder temperature and longitudinal displacement of bearings.A method to finely simulate the longitudinal displacement of bearings is proposed,which takes the non-linear time-varying coefficient regression model as the basis.The longitudinal displacement of bearings computed by this method was compared with the measured data and the simulated data from a multivariable linear regression model.The results show that the one-day longitudinal displacement of bearings exhibits significant non-linear time-varying correlation with the main girder temperature,with related characteristic curves revealing the flat circle traits.The non-linear time-varying coefficient regression model can generate simulation results approximating to the variation trend of measured data,the absolute error between the two is 10.5%,whereas the absolute error between the simulation results from the multivariable linear regression model and the measured data is 30.79%,proving that the former is superior in terms of fitting accuracy.