基于离差和改进的大坝变形监测粗差判别方法

A Gross Error Discriminant Method Based on Difference Squares in Dam Displacement Monitoring

随着大坝监测自动化的普及,在大坝安全分析评估中寻求可靠的粗差判别方法显得越来越重要,而基于回归统计的粗差判别准则中,由于粗差数据也会先参与回归建模,由此导致回归模型本身可能失真,从而易引发后续的误判和漏判粗差。对此,提出了一种针对混凝土重力坝变形监测的粗差判别方法。首先,计算样本中每个测值的离差和;其次,基于离差和剔除部分疑似粗差的测值,从而建立回归模型;最后,根据回归模型计算每个测值的拟合残差,并基于肖维勒准则最终确定粗差。实例仿真表明,该算法不受原始数据粗差的影响,具有更高的精度和鲁棒性。

With the popularization of dam monitoring automation,it is more and more important to find a reliable gross error discriminant method in dam safety analysis and evaluation.In the gross error discriminant criterion based on linear regression,the gross error data will participate in the regression so that the regression model may be distorted and easy to cause subsequent misjudgment.In this paper,a discriminant method for the gross error of concrete dam is proposed.First,each data difference square would be calculated.Secondly,based on the difference square,the suspected gross error data would be excluded to establish a reliable regression model.Finally,each data residual is calculated based on the regression model,and the gross error is determined by Chauvenet criterion.The example shows that the algorithm is not affected by the gross error of the original data,and has higher precision and robustness.