Modified Smith predictor with PID structure for control of multivariable processes

Abstract

Industrial processes with dead time, can be controlled using a model-based control strategy called 'Smith predictor' (SP) to compensate for the dead time part. Most of the control loops in process industries use PID loops. By introducing a tuning parameter (Ksp) in the control law, to switch between SP and proportional-integral-derivative (PID) modes, an SP with PID (SP-PID) is synthesised in this work to counteract the offset and to obtain a better performance during set-point change as well as load changes. Performance of SP-PID to control linear MIMO systems (square, stable, non-singular linear) are carried-out with decentralised scheme by studying set-point change and load disturbances separately. Present results have been compared with similar existing control strategies. Sensitivity and robustness issues of the present control scheme are also discussed.