基于参数收敛的时变参数沉降预测模型

当时变参数沉降预测模型中的参数随时间不收敛时,模型对中长期沉降的预测精度不高,不能预测最终沉降.基于沉降预测模型参数的物理含义，分析模型参数随时间的变化规律,采用随时间收敛的函数来描述模型参数随时间的变化,建立相应的时变参数沉降预测模型.根据前期实测数据,采用最小二乘法对预测模型进行求解.通过引入遗忘因子来加大近期实测数据的权重,进一步提高模型的预测精度.实例分析表明,所提出的时变参数预测模型拟合精度高于现有的沉降预测模型,对全过程的实测沉降数据都具有很好的拟合精度,并对最终沉降做出合理的预测.

Time-dependent parameter model for settlement prediction considering parameter convergence

When the model parameters do not converge, the time-dependent parameter settlement prediction model has low precision for middle- or long-period prediction and cannot predict the final settlement. Based on the model parameters’ physical meanings, the rule of parameter varying with time was analyzed, and a function converged with time was used to describe the parameter, and then the corresponding settlement prediction model was established. The prediction model was solved by the least squares method using the pre-measured data. A forgetting factor was introduced to improve the model prediction precision by increasing the weight of latter observed data. Case study shows that the time-dependent parameter settlement prediction model has higher prediction precision than those of the existing prediction models, and can simulate the whole course of observed settlements and predict the final settlement．