摘要
大坝安全管理已逐步由粗放型向信息化转化,然而传统的误差识别技术识别精度不高,造成大量异常数据进入原始信息库,导致后期分析结论的偏离,成为信息化建设的制约因素。为此,提出了基于稳健估计的大坝监测数据粗差识别方法,即以稳健MM回归替代传统建模方式,同时结合MCD稳健估计确定异常识别阈值,并实现了程序编制。实例应用结果表明,该方法能有效避免由于异常值存在而导致的模型崩溃及阈值失效,且识别更加可靠。
The management of dam safety has gradually changed from the extensive type to informationization.But the traditional error recognition technology has low recognition accuracy,which causes a large amount of abnormal data to involve in the original information library.The deviation of the later analysis conclusions becomes a constrained factor for information construction.For that reason,this paper proposed to replace the traditional modeling method with the MM robust estimation to determine the threshold of abnormal recognition based MCD robust estimation,and to realize the programming.An example results show that the new method can effectively avoid model crashes and threshold failures due to outliers,and the identification is more reliable.
作者
赵泽鹏
陈建康
张瀚
李艳玲
吴震宇
ZHAO Ze-peng;CHEN Jian-kang;ZHANG Han;LI Yan-ling;WU Zhen-yu(State Key Laboratory of Hydraulics and Mountain River Engineering,Sichuan University,Chengdu 610065,China;College of Water Resource & Hydropower,Sichuan University,Chengdu 610065,China)
出处
《水电能源科学》
北大核心
2018年第12期68-71,共4页
Water Resources and Power
