论文部分内容阅读
目前常用的时间维监控模型主要有时间序列模型、回归分析模型、灰色理论模型、卡尔曼滤波模型、泊松生命回旋模型等,模型虽然在某些简单或特定的工程中均可取得了较好的监控效果,但在一些沉降变化复杂的地区,由于模型仅考虑了离散数据的随机性或结构性特征,没有考虑到数据的变化特征和时间相关性,致使模型预测监控的效果总是差强人意。本文研究了基于模拟退火法的Kriging时域模型,其在传统Kriging模型的基础上,引入目前比较成熟的模拟退火法(SAA)对变异函数拟合模型中的参数进行寻优,以提高变异函数模型的精度,使模型能更准确地描述时域变量的变异特征。
At present, time-series model, regression analysis model, gray theory model, Kalman filter model and Poisson life cyclotron model are the most commonly used time-based monitoring models. Although the model can achieve better results in some simple or specific projects However, in some areas with complex settlement changes, the model predictive monitoring effect is always unsatisfactory because the model only considers the randomness or structural features of discrete data, and does not take into account the changing characteristics of the data and the time correlation. In this paper, the Kriging time-domain model based on simulated annealing is studied. Based on the traditional Kriging model, the current simulated annealing method (SAA) is introduced to optimize the parameters of the variation function fitting model to improve the variation function The accuracy of the model allows the model to describe the variability of time-domain variables more accurately.