Transparent Control Based On Constraints Consistency

In process industry, including pharmaceutical industry, papermaking industry, petroleum industry,chemical industry, etc. Model Predictive Control (MPC), as one of the most typical Advanced Process Control (APC) technology, has been found widely applicated. Through the application of MPC, the safety in production of enterprises can be guaranteed, the variation in process has been reduced, the stability of the system has been improved, and enterprises economic benefits enhanced effectively. The increase of economic benefits is closely related to edge control of operating point. Therefore, objective function based on the economic bebefit is employed to obtain the optimal operating point, which is considered as the set point or a reference trajectory of MPC dynamic optimization. In the industrial cascade predictive control system, the function of each layer is different and independent, and the unconsistency of constraints between upper layer and lower layer exists commonly. Usually, the set point is obtained only based on objective of the current layer. It may be unfeasible when transfered to the next layer, even resulting in saturation in controller, which leads to the deterioration of control performance and damage of the economic benefits. To solve these problems, contributions on constraint consistency of transparent predictive functional control system are developed in this thesis, which can be summerized as follows:1. The traditional MPC algorithm for fast servo system may be followed by the control input with unknown law, hence the solving of the quadratic optimization problems cannot be avoided. To overcome this problem, Predictive Functional Control (PFC) algorithm is developed, which makes the input of control variable more regular, reduces the online calculation effectively, and improves the speed of response.2. LQGO would be chosen as steady state optimization in the cascade predictive control system. An equality constraint condition about the relationship between input and output variance is added to ensure that the calculated optimal setpoints based on the steady optimization of upper layer are more practical.3. In order to solve inconsistent constraints, two typical transparent control loop of LQGO—PFC—PID and LQGO—QMPC—PID are developed, and the monotonicity of constraints in the transfer process is proved. The back-calculation is used in a coordinating process of multi-layer structure to ensure constraints consistency between the upper layer and basic control layer by adding constraints of basic control layer to the process of optimization in the upper layer. Then the consistency of constraints in multi-layer MPC control is ensured and multi-layer structure is optimized. To furtherly illustrate the effiectiveness of the proposed approach, the back-calculation is employed under different constraints.