Theory analysis of nonlinear data reconciliation and application to a coking plant

The estimation methodology of process variables usually consists of three parts: classification of process variables, gross error detection, and data reconciliation. In this paper, we proposed a modified M-estimator method for the covariance estimator which depends on the results from robust statistics to reduce the effect of the gross errors. We consider the Lagrange multipliers method and successive linearization method for nonlinear data reconciliation. Finally, the example of a coking plant is presented to illustrate the effectiveness of the revised M-estimator method and nonlinear data reconciliation methods. In this paper, the classifying, estimating, and adjusting of the process variables are based on a components balance and total flow rates balance. The comparative results of the introduced methods are given and demonstrate the successful application of the proposed method to reconcile actual plant data from a complex chemical process.