基于多元线性回归模型的大坝变形预报研究

Forecast for dam deformation based on multiple linear regression model

大坝变形监测预报的准确性对大坝安全评估具有重要作用,而其变形又受诸多相互关系复杂的外界因素的影响。阐述了目前大坝变形预报的几种主要模型和方法,重点对多元线性回归模型进行了分析。针对温度因子和时效因子的特点,在建模时考虑了温度因子对大坝的时延影响,采用三角函数对温度因子进行表述;对时效因子采用衰减蠕变因子描述,建立了科学合理的回归方程。以三峡工程大坝安全监测为例,采用M at-lab语言对回归模型进行了编程实现。计算结果表明,采用多元线性回归模型能有效地对大坝变形进行预报,预报精度较高,对大坝安全评估具有重要意义。

The accuracy of dam deformation forecast has an important role in dam safety evaluation,and the deformation is influenced by complicated multiple-relation factors.We briefly describe several major models and methods for dam deformation forecast and focus on multiple linear regression model.With consideration of the characteristics of temperature factor and aging effect factor,time-delay effect of temperature on dam is considered in model establishment and the temperature factor and aging effect factor are expressed by trigonometric function and damping creep factor respectively.Taking TGP dam safety monitoring as practical case,the computer program for the regression model is realized by Matlab.The calculation results show that multiple linear regression model can effectively forecast dam deformation with higher accuracy and has an important significance for dam safety evaluation.