Integrated data reconciliation with generic model control for the steel pickling process

Abstract

To implement an advanced control algorithm, measurements of process outputs are usually used to determine control action to a process. Nevertheless, measurements of process outputs are often subjected to measuring and signal errors as well as noise. Therefore, in this work, Generic Model Control (GMC), an advanced control technique, with data reconciliation technique has been applied to control the pH of the pickling process consisting of three pickling and three rinsing baths. Here, the data reconciliation problem involves six nodes and fourteen streams. The presence of errors in the data set is determined and identified via measurement test, In addition, the measurement error covariance is initially assumed to be a known variance matrix and is updated every iteration. Simulation results have shown that the reconciled process data give a better view of the true states of the process than raw measuring data. With these reconciled process data, the GMC controller can control the process at a desired set point with great success.