流程工业中显著误差检测算法及应用研究

论文编号:ZD987    论文字数:10011,页数:32

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。且本文通过测量检测法和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。

关键词：显著误差检测、测量检测法、F统计量

Abstract
 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 t hat t he data reconciliation procedure tends to spread the gross errors overall t he 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 and this paper proposed a new test method based on creating measurement test and F-statistic for detecting gross errors.
Keywords：gross error detection、measurement test、F-statistic

目 录

摘要……………………………………………………………………………………II
1 概论………………………………………………………………………………………1
2 测量检测法仿真………………………………………………………………2
2.1  显著误差检测………………………………………………………………………2
2.2  传统数据协调模型………………………………………………………  3
2.3  测量检测法算法……………………………………………………………………4
2.4  仿真研究……………………………………………………………………………5
3  F统计量建立……………………………………………………………………11
 3.1  方法描述………………………………………………………12
 3.2  F检测法………………………………………………………………………………15
4  F统计法仿真…………………………………………………………………………16
 4.1  仿真1……………………………………………………………………16
 4.2  仿真2……………………………………………………………………19
5   总结…………………………………………………………………………………27
致谢………………………………………………………………………………………28
参考文献…………………………………………………………………………………29