Performance Analysis And Improvement Of Model Predictive Control Algorithms

The research and application object of Model Predictive Control(MPC)is a control problem with optimization requirements.In practice,almost all processes have constraints.MPC's greatest advantage lies in its ability to handle complex constraints.Another advantage of MPC is that it can be used in MIMO system.Based on this advantage,different methods can be used.Although the solutions are different,the essence of MPC is the same,that is,to optimize the manipulated input in order to predict future behavior in the control process.Of course,MPC can be applied to systems with different performance characteristics in different fields,such as continuous-time systems,discrete-time systems,linear systems,non-linear systems,uncertain systems and so on.According to the research characteristics of different systems,adding MPC algorithm which accords with one or more performance indicators and improving the existing algorithm is the focus of this paper.In this paper,according to the different characteristics of the control system,different pertinent MPC algorithm is used and improved.Then the dynamic and steady-state performance characteristics of the control system,such as stability,robustness,convergence,are analyzed.The feasibility and effectiveness of the improved algorithm are verified by simulation experiments.The main contents are as follows:(1)The research of model predictive control is usually based on discrete-time systems,and most control systems are continuous in practice.Therefore,the stability and robustness of MPC in continuous-time systems are studied.(2)With the widespread existence of hard constraints in the actual industrial production process,although there are many control schemes,there are few methods with good control performance.Therefore,the effectiveness and stability of a shrinking SM-MPC algorithm are studied in a specific hard-constrained discrete system.(3)In recent years,MPC has been successfully applied to many slow dynamic systems with process input and output constraints.However,for systems with short sampling intervals,due to the large amount of online computation and the complexity of the algorithm,MPC is relatively limited.Therefore,in view of this challenge,a robust self-triggering MPC method is proposed for linear systems with constraints,which combines self-triggering MPC with robust control.(4)In the case of non-linear MPC,the optimal control problem is usually researched based on non-convex non-linear programmingsystem(NLPS),which can not guarantee the global(or even local)optimal solution in sampling time.Therefore,the stability condition of suboptimal MPC is of great significance for practical application.Suboptimal MPC is a fast control algorithm for optimal control problems by using suboptimal solutions.Suboptimal MPC has inherent robustness to systems with soft terminal region constraints and extends to systems with hard terminal region constraints.