Steady-State Target Optimization Strategy In Model Predictive Control And Its Application

Model predictive control (MPC) is a model based computer control algorithm which is presented in 1970s. Since its presentation, it has been widely used in process industry especially in refinery and petrochemical processes. One of the major reasons for this success is that the strategy can easily be used to solve multivariable input and output constrains. Recently, along with the scale of commercial enterprises expanding continuously, the plant become more and more complex which leads to many constrains. However, process limitation and constraints may perturb the plant from their desired working point, and the disturbance entering the plant or new input information from the operator may change the location of the optimal steady state. So how to deal with the optimization and control of the complex system in order to make system running in a more economic state is the main problem of this research. In this thesis, we studied the problem from three aspects which are as follows.1. MPC calculation is separated into steady-state and dynamic optimization. Considering the effects of measured disturbance, the steady-state objective is recalculated at each sampling time. The difference between the model prediction and the measured output at the current time is introduced into the model to incorporate feedback and the velocity constraints in steady-state target calculation ensure the steady-state objective is compatible with the velocity constraints in dynamic MPC optimization. Finally, this control strategy has been successfully applied to the control of the Shell heavy oil fractionators benchmark problem.2. The soft constraints adjustment and target relaxation of steady-state target calculation are systemically investigated. The reason for requiring soft constraint adjustment and target relaxation simultaneously is that the result is not satisfactory when the optimization problem unfeasible or the feasible region is apart from the desired working point due to system dynamic. The goal priority factor is introduced to describe the priority of the constraints and targets, thereby the soft constraints adjustment and targets coordination are solved systemically in real-time steady-state target calculation. Simulation is processed with the example of the shell heavy oil fractionators benchmark problem, and the result shows the validity of the proposed algorithm. 3. A granulation system presented by Pottman et al. is used to demonstrate the proposed algorithm. The closed-loop simulation for the granulation system using the steady-state target calculation strategy based on goal programming shows the validity the proposed algorithm. The system can run in a more economic state.