Research Of Zone Predictive Control Based On MLD

Predictive control has become a widely used algorithm in dealing with the constrained multivariable optimal control problem in the process industry.Because of disturbance,model mismatch and other factors,infeasible problems can be met in constrained optimization and thus control target cannot be reached exactly.In order to solve these problems,soft constraint and zone control schemes have been adopted.The conventional zone control scheme converts zone control objectives into system constraints directly,and has some disadvantages,such as small feasible set,slow response and poor robustness.According to these problems,this paper investigated a new hybrid zone predictive control algorithm based on mixed logic dynamic(MLD)model.The algorithm introduces logical variables to represent the deviation region of controlled variable and integrates the logic rule into algorithm formulation.This algorithm can make the controlled variables return to the expected zone faster;By introducing 1-norm cost function in hybrid predictive control algorithm,computational efficiency of the algorithm has increased;Boundary funnel technology is introduced in the hybrid zone predictive algorithm to prevent overlarge transient deviation and improve the robustness of zone predictive control.Algorithms in this paper are simulated on Shell tower,and achieve satisfactory results.The main works in this paper are as follows:The zone predictive control based on MLD model is studied,which introduces the logical variables into the design of controller.The objective function base on logical proposition is integrated into the framework of the predictive control optimization in the form of hybrid logical inequality constraints.Then logical variables are introduced to represent the deviation region of controlled variable,by using the value of the logical variable to detect the deviation area of the controlled variable,the corresponding error weights are set.This weighted zone predictive control algorithm based on MLD can make the controlled variables return to the expected zone faster.In order to increase computational efficiency of the hybrid zone predictive control,hybrid zone predictive control based on linear programming is proposed.The performance index takes 1 norm and the optimization problem is converted into MILP,avoids solving MIQP problem.Computational efficiency is increased.By introducing boundary funnel constraints,this paper studies the hybrid zone predictive control with boundary funnel constraints.On the basis of zone predictive control,in order to prevent loss of the stability of the system from the drastically change of process variable,the boundary funnel constraints were added to the controlled variables in the period of predictive horizon.By introducing the logical variables and integrating the logical rules into the solution of algorithm,switch of the object and constraint conditions are realized.Overlarge transient deviation is prevented and robustness of zone predictive control is improved.Simulation has been carried out in the Shell tower process for the algorithms in this thesis.Simulation results show the effectiveness of the proposed algorithms.