中文摘要：

建立水工建筑物安全监控模型是预测和控制大坝的运行、避免重大事故发生的不可或缺的重要手段，而大坝安全监测数据中经常出现的异常值对监测模型的稳定性和精度产生较大影响，针对这个异常值问题，采用鲁棒回归方法，有效地剔除或削弱了监测数据中的异常值的影响，实例分析表明：在监测数据中含有异常数据的情况下，基于鲁棒回归的大坝安全监测模型的稳定性和预测效果能够得到保证。

英文摘要：

Hydraulic structure safety monitoring modeling is an important means to predict and control dam operation and prevent major accident. But the abnormal values in the monitoring often influence the stability and precision of monitoring model. The robust regression is introduced to eliminate or weaken the influence of abnormal value. The analysis results of a case study show that the stability and prediction effect of dam safety monitoring model can be ensured by using robust regression under the condition of abnormal values.