数据协调中测量数据方差–协方差估计及过失误差序列补偿算法

数据协调测量误差的方差-协方差矩阵(也称Q矩阵)通常是由操作人员根据仪表的精度事先给定的,由于没有考虑仪表精度的变化,很可能会造成数据的不一致或不准确.基于空间冗余的约束残差,本文提出了一种测量方差-协方差Q矩阵的估计方法,有效地减少了对Q矩阵先验知识的依赖.针对常见非线性或双线性问题,提出了估计方法的应用方案.并给出了侦破过失误差的序列补偿法.最后,某焦化厂的应用示例表明.该方法对于获得Q矩阵的最初估计和侦破过失误差是有效的.

Available estimation of measurement error variance/covariance and sequential compensating algorithm of gross error for data reconciliation

Because of the ignorance of the variation in instrument precision,the variance or covariance(matrix Q) of measurement error for data reconciliation given by an operator often causes inconsistency or inaccuracy in data rec- onciliation.Based on the constraint residual of spatial redundancy,a method for estimating the covariance matrix Q of measurement error is proposed which effectively reduces the dependency on a priori knowledge of matrix Q.The applica- tion of this estimation method to nonlinear or bilinear systems is described.The sequential compensation algorithm is also presented to detect the gross error.Finally,the application results from a coking sub-plant show that this method is feasible in obtaining the initial estimates of Q and detecting the gross error.