Decentralized Robust Model Predictive Control For Uncertain Large-scale Systems

The uncertain large-scale systems are coupled by several subsystems.Uncertainty is common in the actual system.With the development of society,the actual production process not only requires association terms between large-scale systems,but also needs to increase or reduce the related terms.Therefore,the control problem of uncertain interconnected large-scale systems is worth studying.For the system's uncertainty,robust predictive control can control the performance index of the actual system within the acceptable range and ensure the robust stability of the closed-loop system.For the large-scale and integrated system scale,decentralized control can guarantee the normal operation of the large-scale systems,and the system can operate normally for production and manufacture.Predictive control can solve the problem of information shortage caused by decentralized control.Decentralized predictive control is an effective strategy for large-scale systems.The research of the early large-scale systems decentralized predictive control is focused on the representation and classification of large-scale systems interconnections,but the system uncertainty is not considered.The decentralized predictive control of the large-scale systems is further developed.The Plug-and-Play(PnP)control is proposed for the changes of the large-scale systems association terms,which is very practical for the actual industrial process.It is widely used in electric power system,supermarket refrigeration system and so on,so it is an important research field.In this paper,a decentralized robust predictive controller with stability assurance for large-scale systems with uncertain parameters and value bounded uncertainty is presented for uncertain large-scale interconnected systems.The main contents are:For large-scale systems with uncertain parameters,the design of its decentralized robust predictive controller is invested.The parameter uncertainty is assumed to satisfy the limited conditions for large-scale systems.For the related terms,it is considered as external disturbance and does not consider collective quantities,and the state of the system is tracked with the state of the nominal system.The finite time domain optimization problem is transformed into a linear matrix inequality(LMI)constrainproblem.A decentralized robust predictive control algorithm based on Plug-and-Play control is proposed for large-scale interconnected systems with value bounded uncertainty.The value bounded uncertainties are converted into linear matrix inequalities(LMI)through min-max optimization problems.The interconnected terms between the subsystems are decomposed through the block diagonal matrices and then obtained.PnP operation is used to get the influence of the change of correlation on the stability of the system.