ISSN 1671-1092 CN 33-1260/TK
大坝与安全 ›› 2026 ›› Issue (2): 37-.
随着数字孪生水利工程建设不断推进，安全监测作为数据底板构建的重点，是后续专业模型和业务应用的基础。然而，采集和传输过程中产生的异常数据影响了数据可靠性和完整性，因此识别和清除异常数据是提高监测数据质量、确保大坝安全评估有效性的关键。本研究采用广义相加模型（GAM）结合孤立森林算法（IF）进行大坝监测异常数据识别。首先通过IF识别监测数据中的明显异常值，在此基础上利用GAM提取监测数据序列的趋势项，对剔除趋势项信息的残余分量采用3δ 方法进行精细化异常值识别，实例数据分析表明该方法可以有效识别大坝监测数据中的异常值。
With the continuous construction of digital twin hydraulic engineering, safety monitoring, as the focus of data base construction, is the basis of subsequent professional models and business applica_ tions. As the abnormal data generated during the collection and transmission process would affect the re_ liability and integrity of data, identifying and clearing abnormal data is the key to improve the quality of monitoring data and ensure dam safety assessment. Therefore, the generalized additive model combined with isolated forest algorithm is applied in the identification of abnormal data of dam monitoring. Firstly, the obvious outliers in monitoring data are identified by isolated forest algorithm. Then, the trend items of monitoring data series are extracted by generalized addition model, and trend item information is elim_ inated. And the 3δ method is used for refined identificative of outliers. It is proved that the method can effectively identify the outliers in dam monitoring data through the example data analysis.
摘要： 随着数字孪生水利工程建设不断推进，安全监测作为数据底板构建的重点，是后续专业模型和业务应用的基础。然而，采集和传输过程中产生的异常数据影响了数据可靠性和完整性，因此识别和清除异常数据是提高监测数据质量、确保大坝安全评估有效性的关键。本研究采用广义相加模型（GAM）结合孤立森林算法（IF）进行大坝监测异常数据识别。首先通过IF识别监测数据中的明显异常值，在此基础上利用GAM提取监测数据序列的趋势项，对剔除趋势项信息的残余分量采用3δ 方法进行精细化异常值识别，实例数据分析表明该方法可以有效识别大坝监测数据中的异常值。
中图分类号:
