Generalized predictive control with extended predictive conyrol for two-tank level control via distributed control system

Abstract

Predictive control algorithms emerged in the middle of the seventies from the side of industrial applications. In this thesis, a new predictive control termed extended predictive control (EPC) is implemented on the two-tank level control to control the water level at the middle operation point of each tank via distributed control system (DCS). First the original generalized predictive control (GPC) testing is performed on the plant. The response indicated a slow response of process. On the other hand, the requirement of fast closed-loop response, minimum overshoot with well conditionality is important to the level control industrial for reducing operation setup time. From the results of generalized predictive control, EPC is studied and practically implemented. EPC drastically improved the system matrix conditionality by minimizing condition number and increasing determinant of the system matrix to evaluate a single tuning parameter. First the control performance of EPC and GPC is tested by MATLAB programming for middle setpoint trajectory of the nominal model. The results showed that the EPC provided good control of water levels with faster closed-loop response and minimal overshoot for the same weighted matrix condition.