Impact of Model-Plant Mismatch to Minimum Variance Benchmark in Control Performance Assessment

Abstract

Control Performance Assessment (CPA), since its first developed, has caused great attentions and been widely used in control systems. Although a large number of methodologies have been implemented, performance assessment index, as an economic and common method to monitor the safety and operation condition of the controlled device, is computed under an assumption that models supervised in the control loop systems are steady and would not change during a sufficient period. However, the existence of Model-Plant Mismatch (MPM) could cause large turbulence and fluctuation in closed-loop and affect the benchmark which is fit to evaluate the control performance. Hence, it is necessary to analyze the impact of Model-Plant Mismatch on control performance assessment benchmarks. In this paper, we analyze the possible changes caused by MPM towards performance criteria in industrial process with time delay. Considering that many efforts in the domain of Model-Plant Mismatch are focused on true plant process without taking the time delay mismatch into consideration, we fulfill these tasks and distinguish the impacts of plant mismatch as well as that about time delay one. The plant or the time delay mismatches are simultaneously discussed to illustrate the disadvantageous impacts towards CPA. Finally we got the conclusion that the Model-Plant Mismatch will contribute to the alterations in real control loop output variances. But only the changes in time delay coefficient will directly affect the Minimum Variance (MV) benchmark and leads to the changes in both the system output variances and the Minimum Variance index.