Analysis of multivariable control performance assessment techniques

摘要: Abstract Control performance assessment (CPA) techniques provide an indication of how current controller compares with what would be considered to ideal. The ideal is typically referred as a ‘benchmark’. This paper argues that there are two fundamental requirements for any CPA algorithm. first it should able detect change in the control system and second identify potential improvement can made if were re-tuned or re-designed. ability multivariable address these issues reviewed novel monitoring strategy application systems proposed. this improved approach detecting changes identifying improvements possible through re-tuning illustrated using simulated industrial data.