The Economic Performance Assessment Of MPC Based On Optimal LQG Benchmark

Model predictive control now as the most popular control strategy has been used in more and more industry control systems. However, for some outer and inner uncertainty, the system operating points may go far away from the optimal set-points. In order to judge whether the system is in good running condition and economical or not, the economic performance assessment of control system is proposed. The main idea of economic performance assessment is that by assessing the system’s economic performance, the optimal set-point of MPC and also the best economic performance of the system can be obtained. Then some corresponding adjustment strategies are given to improve the system economic performance.The main content of this paper is organized as follows:First, the traditional LQG benchmark is unbalanced distributed, thus leading a bad regression results. In order to get a more effective LQG benchmark, the traditional LQG benchmark is modified with an index weight coefficient. And finally the accuracy of the method is verified through a simulation case.Then, as for multivariate control system, the accurate LQG benchmark is usually hard to get. In this work a method is propose to get the LQG trade-off curve for MIMO system. The LQG trade-off curve is more intuitive and with more practical significance by this method. And a simulation is given to verify the feasibility and effectiveness of this idea.Eventually, the economic performance assessment with the proposed modified LQG benchmark is established. In this work, a user-defined "back-off strategy is selected to confront the uncertainty acting on the system, then the economic performance assessment is translated into a steady-state stochastic optimization. By solving the optimization problem, the system optimal set-point and best economic performance are obtained. Then some corresponding strategies for controller tunning are given to get a better economic performance. And finally simulations are given to verify the effectiveness of proposed method.