Indian Journal of Science and Technology
DOI: 10.17485/ijst/2017/v10i7/111454
Year: 2017, Volume: 10, Issue: 7, Pages: 1-5
Original Article
Truong Thanh Tuan, Lemma Dendena Tufa, Mohamed Ibrahim Abdul Mutalib and Kelani Rasheed Olakunle
Department of Chemical Engineering, Universiti Teknologi Petronas, 32160 Tronoh, Perak, Malaysia; [email protected], [email protected], [email protected], [email protected]
Objectives: In advanced process control, especially model predictive control (MPC), a model is needed to calculate the input (manipulated variable) to the plant to track the set-point. The model used for MPC is usually empirical model, usually a state space model in the open literature, identified by system identification. The problem is that the empirical model is never completely accurate to represent the plant, a reason that brings about an offset in set-point tracking by MPC. In addition, in the presence of disturbance, the accuracy becomes much worse. Method: In this work, we recommend state estimation for the state prediction according to measured output at each iteration calculation to obtain an equal output prediction with output measurement from the plant. Findings: It was found out that integrating MPC and Kalman filter could facilitate linear offset free MPC. Application: The success of this approach is demonstrated using an integrated MPC and Kalman filter in SimulinkMatlab to control the dynamic Depropanizer process in Hysys.
Keywords: Kalman Filter,Linear Offset Free MPC, Matlab-Hysys Interface
Subscribe now for latest articles and news.