摘要
传统监测资料分析是在假定监测数据服从正态分布的前提下,采用经典最小二乘法进行建模分析的,但最小二乘法善于吸收异常值,不具备抗差能力。提出应用抗差最小二乘法建模,并对其计算方法以及相关参数进行了系统的研究,提出了监控模型的优化方案。同时通过监测实例证明,抗差最小二乘法计算模型优于最小二乘法计算模型,能够降权使用可疑值,淘汰异常值,具有较强的抗差能力,预测精度高。
Conventional monitoring data analysis is conducted by least square method under a presumption that monitoring data obeys with the normal distribution,however the least square method absorbs abnormal values,which doesn＇t have robustness.We propose the robustness square method and study its calculation method and relevant parameters.And the optimized scheme of monitoring model is suggested.The monitoring example proves that the model established by the robustness estimation method is superior to that by the least square method;the robustness estimation method can reduce the weight of suspicious value and eliminate abnormal value,therefore,it possesses high robustness with high accuracy.
出处
《人民长江》
北大核心
2011年第9期94-97,共4页
Yangtze River
关键词
抗差最小二乘法
权函数
统计模型
预测精度
大坝安全监测
robustness least square method
weight function
statistical model
prediction precision
safety monitoring of dam
