Event-Triggered Model Predictive Control Of Networked Systems

Model predictive control(MPC)is currently widely utilized in industrial control systems for its optimization-based nature and capability in explicitly handling system constraints.Accompanied by the rapid development of communication and signal processing technology,networks play more and more important roles in modern industrial systems,and many traditional industrial systems have been upgraded and transformed into networked control systems(NCS).Considering the fact that network resources are quite scare in large scale NCS,traditional MPC that periodically performs optimization and broadcasts input signals may be not suitable.Therefore,event-triggered MPC(ETMPC)algorithms are needed to make the controller resource-aware.Motivated by this,this dissertation is dedicated to the event-triggered MPC of systems with state and control constraints.Specifically,the following cases are addressed: constrained linear and nonlinear systems with additive disturbances,constrained nonlinear systems with additive disturbances and parametric uncertainties,and dynamically coupled nonlinear systems with input constraints.Based on the linear interpolation technique,min-max optimization,dynamic programming,invariant set theory,and Lyapunov stability theory,four schemes of ETMPC are developed with feasibility and stability guarantees.The main results and contributions are four-fold:1.A dynamic ETMPC scheme is developed for constrained linear systems subject to additive disturbances.By interpolating a set of stabilizing linear feedback gains,a feedback policy is constructed to significantly enlarge the feasible region while remaining computationally efficient.The cost function is designed to penalize the weighting factors associated with the low-gain feedback laws to optimize the policy online.To achieve robust constraint satisfaction,the system constraint sets are properly tightened according to the interpolating coefficients.Furthermore,a dynamic event trigger also characterized by the interpolating coefficient is proposed,based on which the optimization problem is only solved at sporadic triggering time instants.The recursive feasibility of the proposed algorithm and stability of the closed-loop system are proved.2.An ETMPC algorithm is developed for constrained nonlinear systems with additive disturbances.First,a time-varying tightened state constraint is computed according to the Lipschitz constant of the system dynamics and the disturbance bound to fulfill robust constraint satisfaction.Then an event-triggered scheduling strategy is designed;it monitors the gap between the predicted system state and the actual state to determine the triggering time instant.The lower and upper bounds of the triggering interval are derived,implying that the Zeno behavior is avoided.The recursive feasibility of the algorithm and robust stability of the closed-loop system can be ensured with the proposed algorithm.3.A robust self-triggered MPC strategy is proposed for constrained nonlinear systems subject to parametric uncertainties and additive disturbances,and by employing the affine disturbance feedback scheme,extensions of the proposed method are made for linear systems with improved computational efficiency.In the developed method,we take advantage of the min-max MPC framework to consider the worst case of all possible uncertainty realizations to achieve robust constraint satisfaction.The cost function is designed based on which a self-triggered strategy is introduced via optimization.The conditions on ensuring algorithm feasibility and closed-loop stability are developed.In addition,we show that,for linear systems with disturbance feedback,the main feasibility and stability conditions reduce to a linear matrix inequality.4.The distributed ETMPC problem of dynamically coupled nonlinear systems with input constraints is studied.In the proposed scheme,each agent constructs its local model by taking advantage of the asynchronous communication among agents.A constraint is added to the local optimization problem to limit the change of the predicted states at sequential triggering time instants.The distributed event-triggering rule is realized by locally comparing the gap between the predicted state and the actual state with a threshold.Thanks to them,the mutual disturbances caused by dynamical coupling are proved to be bounded,and the Zeno behavior is avoided.Finally,the sufficient conditions ensuring algorithm feasibility and closed-loop stability are provided,facilitating the practical use of the algorithm.At the end of the dissertation,the main contributions are summarized and the problems to be explored in the future are discussed.