Research On Control Strategy Of Periodic Disturbance Rejection

Since been developed in the 70s of last century, Model predictive control (MPC) has become a very important sub-discipline with rich theory foundation and practical application. In spite of the widespread uses on process control industry, it didn't have a specific strategy for periodic disturbance rejection yet. Beside, few study works has been carried out in this research field. Some control strategies of periodic disturbance rejection which based on MPC are researched in this dissertation, and the main research works are as follows:(1) Research progresses of disturbance rejection in MPC has been reviewed and summed up. Relevant control methods with periodic disturbance rejection have been introduced and categorized.(2) The shortage of MPC in periodic disturbance rejecting has been analyzed and illustrated by simulations. An improved MPC algorithm based on repetitive control method is derived; the effectiveness has been demonstrated by simulation results. And the limitation of this method has also been intensively studied.(3) An iterative learning model predictive control algorithm that uses the optimal law has been proposed. A learning control loop is adopted based on the feedback loop in the traditional control systems. The periodic disturbance rejection is handled by ILC while MPC doing its job of system process control. It can achieve a better control performance while period mismatch appears between disturbance signal and the controller.(4) In order to eliminate the influences from the impenetrable periodic disturbances which are common in industrial process, a model predictive control algorithm with online model identification part of periodic disturbance is introduced. Meanwhile, parameter selection principles of the algorithm have been analyzed in consider of periodic disturbance characteristics. It can also obtain a great performance with variable-frequency disturbance problems in actual process, which covers the main shortage in other control strategies and has a good practice value.