摘要
为研究监测数据异常值对大坝变形的不利影响,建立一种支持向量机和协方差矩阵(SVM-MCD)耦合的数学模型,采用理论分析的研究方法,基于某实际重力坝和拱坝,对大坝变形数据异常值进行识别。结果表明,通过与传统回归模型对比发现,SVM模型拟合值与实测值更加吻合,表明SVM-MCD模型的正确性;本文模型的计算误差要远小于回归模型,能够更为有效反映大坝变形与各影响因素之间的非线性关系;大坝变形异常值与水位、温度没有明显联系,仅为曲线某处的突变值,符合异常值的分布规律。
In order to study the adverse effects of abnormal values of monitoring data on dam deformation,a mathematical model coupled with support vector machine and covariance matrix(SVM-MCD)was established in this study.For arch dams,outliers in dam deformation data were identified.The results show that by comparing with the traditional regression model,it is found that the fitted value of the SVM model is more consistent with the measured value,indicating the correctness of the SVM-MCD model;the calculation error of the model in this paper is much smaller than that of the regression model,which can reflect more effectively the nonlinear relationship between dam deformation and various influencing factors;the abnormal value of dam deformation has no obvious relationship with water level and temperature,but is only a sudden change somewhere in the curve,which is in line with the distribution law of abnormal value.
作者
王军玉
WANG Jun-yu(Pingliang City Hengsheng Water Conservancy and Hydropower Project Co.Ltd.,Pingliang 744000,Gansu,China)
出处
《水利科技与经济》
2023年第3期154-158,共5页
Water Conservancy Science and Technology and Economy
