Computationally Efficient Aggregation Optimization Strategy of Model Predictive Control

摘要

Model Predictive Control (MPC) is a popular technique and has been successfully used in various industrial applications. However, the big drawback of MPC involved in the formidable on line computational effort limits its applicability to relatively slow and/or small processes with a moderate number of inputs. This paper develops an aggregation optimization strategy for MPC that can improve the computational efficiency of MPC. For the regulation problem, an input decaying aggregation optimization algorithm is presented by aggregating all the original optimized variables on control horizon with the decaying sequence in respect of the current control action.