Performance diagnosis of MPC with model-plant mismatch

摘要: Model predictive controllers have been widely applied in refinery and chemical plant. The process model plays an important role the performance assessment of control. In this work, a methodology is adopted for diagnosis model-plant mismatch (MPM) based on closed-loop operating data DMC system. problem MPM transformed into effectiveness model. Also apply statistic inference to detect occurrence mismatch. This measure successfully illustrated through implementation Wood-Berry tower with three cases show feasibility method.