Design And Analysis Of Model Predictive Controllers Based On Aggregation Strategy

The aggregation strategy of inputs can reduce the burden of online computation of model predictive controller efficiently. However, it brings new problems to stability and impairs the performance of controller due to the additional constraints of inputs. Analyze and generalize the methods of designing classical aggregation matrix by qualitative analysis through simulation. Based on the analysis of attenuation aggregation strategy, propose a new aggregation strategy which can improve the stability and performance of controller, the method of computing aggregation matrix under new aggregation form has been given. The design methods of stable model predictive control cannot be used to the aggregation form. To ensure the stability of MPC based on aggregation form, proposed two new aggregation strategies which named vary-matrix aggregation and with-addition aggregation. The two new aggregation strategy change the aggregation form in every control time to ensure the feasibility of the construct solution in time k + 1, and then prove the stability of the system. Do further research of stability of MPC based on the above two aggregation strategy. As the aggregation strategy influences the controller in two aspects: computation burden and control performance, define the comparison of aggregation matrix based on the two aspects. Obtain the basic methods of comparing aggregation matrix. Analyze the performance of controller based on the two new aggregation strategy proposed by this dissertation.