State Estimation of an Acid Gas Removal (AGR) Plant as Part of an Integrated Gasification Combined Cycle (IGCC) Plant with CO_2 Capture

摘要

An accurate estimation of process state variables not only can increase the effectiveness and reliability ofprocess measurement technology, but can also enhance plant efficiency, improve control systemperformance, and increase plant availability. Future integrated gasification combined cycle (IGCC) powerplants with CO_2 capture will have to satisfy stricter operational and environmental constraints. To operatethe IGCC plant without violating stringent environmental emission standards requires accurate estimationof the relevant process state variables, outputs, and disturbances. Unfortunately, a number of theseprocess variables cannot be measured at all, while some of them can be measured, but with low precision,low reliability, or low signal-to-noise ratio. As a result, accurate estimation of the process variables is ofgreat importance to avoid the inherent difficulties associated with the inaccuracy of the data. Motivatedby this, the current paper focuses on the state estimation of an acid gas removal (AGR) process as part ofan IGCC plant with CO_2 capture. This process has extensive heat and mass integration and therefore isvery suitable for testing the efficiency of the designed estimators in the presence of complex interactionsbetween process variables. The traditional Kalman filter (KF) (Kalman, 1960) algorithm has been used asa state estimator which resembles that of a predictor-corrector algorithm for solving numerical problems.In traditional KF implementation, good guesses for the process noise covariance matrix (Q) and themeasurement noise covariance matrix (R) are required to obtain satisfactory filter performance. However,in the real world, these matrices are unknown and it is difficult to generate good guesses for them. In thispaper, use of an adaptive KF will be presented that adapts Q and R at every time step of the algorithm.Results show that very accurate estimations of the desired process states, outputs or disturbances can beachieved by using the adaptive KF.