测量显著误差的F检验法  

F-Test Method for Gross Errors Detection

作　　者：蔡伟建[1] 李津蓉[1] 

机构地区：[1]浙江科技学院自动化与电气工程学院,杭州310012

出　　处：《传感技术学报》2006年第4期1171-1175,共5页Chinese Journal of Sensors and Actuators

摘　　要：显著误差检测是数据校正技术中必不可少的一环,以往的显著误差检测方法绝大多数都是基于测量残差和约束残差这两个统计量展开研究的。基于测量残差的检测方法首先需要对测量数据进行数据协调,这就会将显著误差分散到各个测值中去,从而会对显著误差的位置做出错误的判断。基于约束残差的检测方法只能对节点的平衡性进行判断,而无法确定显著误差的具体发生位置。为此,本文通过构造一个基于测量值比例关系的F统计量,并与约束残差统计量相结合,对稳态过程中出现的显著误差进行检测。通过对工业数据的仿真结果表明此方法对显著误差十分敏感,其各项性能指标均符合实际工业要求,具有较高的可信度和可应用性。Gross errors must be detected and either deleted of corrected before carrying out the reconciliation. The proposed gross errors detection methods before mostly based on the measurement residual or constraint residual. The main problem with the test based on measurement residual is that the data reconciliation procedure tends to spread the gross errors overall the measurements, thereby failing the test for outliers. Using the methods based on constraint residual is capable to tell which node is imbalance only but can not identify where the gross error is on. In order to avoid these problems, this paper proposed a new test method based on creating a F-statistic for detecting gross errors. The new method has combined the F-statistic with constraint residual statistical to detect gross errors in steady state processes. The result of simulation shows that the new method is very sensitive to presence of gross errors and has a great probability of correctly finding one or several gross errors.

关 键 词：显著误差检测 过程数据 稳态过程 F统计量 比值 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置]