一种用于显著误差检测的改进型NT-MT算法

工业过程中数据的可靠性和准确性是控制过程和管理优化重要基础,但是由于测量过程中不可避免的存在着各种误差,测量值就不能准确的反应实际的化工过程。本文讨论了数据校正对工业过程的主要影响,如果有显著误差的存在,仍按照标准偏差来调整测量数据让其满足物料和能量平衡,所有的测量值将会受到显著误差影响而使得到的数据不能正确的揭示工业过程的状态。本文主要通过对几种传统检测显著误差的分析比较,提出了一种基于MT-NT测量方法的改进型NT-MT检测方法,通过改变测量方差的值来减少含有显著误差测量值所占的权重。仿真结果表明,本论文提出的方法能够更有效地检测和诊断显著误差。

An Improved Approach of NT-MT for Gross Error Detection

Reliable and accurate data is an important foundation of process control and optimization of management.However,because the measurement errors will inevitably exist,the measured value does not correctly reflect the actual chemical process.Data reconciliation and gross error detection are methods intended to deal with instrument measurements that are inconsistent with mass and energy balances.If gross errors exist in process data,all measurements are affected by such biases,and would not accurately indicate the state of the process.In this paper several traditional detection methods are analyzed.An improved method based on the MT-NT measurement is proposed.By changing the value of measurement variance,the weight of measurement value that contains significant gross errors will be decreased.The simulation shows that gross errors can be detected more effectively.