用于含过失误差数据稳态检测的改进滤波法

滤波法是稳态检测应用较广泛的方法 ,但它不适用于检测变量含有过失误差的情况 ,而实际过程测量变量常含有过失误差。该文对滤波法进行改进 ,用新假设代替原不合理假设 ,根据数理统计理论 ,推导出用于稳态检测的统计量。该统计量能应用于检测数据含过失误差的情况 ,且当数据不含过失误差时 ,可以推导出原统计量。数值实验结果显示 ,原方法用于检测数据含过失误差时稳态检测失效 ,而改进后的方法依然能够给出满意的结果

Improved filtering method for steady state tests with measurements containing gross errors

Filtering methods for checking steady state processes can not give correct results when the measurements have gross errors, which often occurs in chemical processing. A new method was developed using probability and statistics theory to develop a statistical filtering variable for steady state tests. The filter is applicable to steady state tests with process data containing gross errors. Previous filtering methods can be derived from the new method for process data without gross errors. Numerical experiments show that the new method reliably provides correct results when measurements have gross errors, while the old method fails to identify steady state.