Modeling and model predictive control of a nonlinear hydraulic system

Abstract

This paper deals with modeling and control of a hydraulic three-tank system. A process of creating a computer model in the MATLAB/Simulink environment is described, and optimal PID (proportional–integral–derivative) and model predictive controllers are proposed. The modeling starts with the creation of an initial mathematical model based on a first-principles approach. Then, the initial model is enhanced to obtain better correspondence with a real-time system, and parameters of the modified system are identified from measurements. The real-time system contains nonlinearities which cannot be neglected, and therefore are identified and included in the final mathematical model. The resulting model is used for a control design. As the real-time system has long time constants, usage of the Simulink model dramatically speeds up the design process. Optimal PID and model predictive controllers (MPC) are proposed and compared. Model predictive controllers for both a linearized model and the nonlinear system are proposed. The techniques described are not limited to one particular modeling problem, but can be used as an illustrative example for the modeling of many technological processes.