稳态过程的数据校正和大误差侦破方法的集成研究

针对实际数据校正问题中同时含有未测变量和带大误差已测变量的情况，提出利用两步投影矩阵方法来进行数据分类，先将稳态模型变换为统计检验所需要的标准形式（只含已测变量），用广义似然比方法侦破出大误差并予以补偿后再对数据校正问题进行求解计算，从而为数据校正及大误差侦破的集成提出了一种通用的策略，并且减少了计算量。

STUDY ON THE INTEGRATED DATA RECONCILIATION AND GROSS ERROR IDENTIFICATION IN STEADY STATE PROCESS

According to the fact that the unmeasured variables and the measured variables with gross errors sometimes simultaneously exist in the steady state process, this paper presents a procedure, in which the variables are classified by two step projection matrix method and at the same time the general steady state model is transformed into the standard form required by statistical test (only measured variables present) , the gross errors are detected and compensated by GLR method and then the data reconciliation is performed. Thus, an integrated method of data reconciliation and gross error identification is developed and the computation is reduced.