Robust Model Predictive Control Based On Communication Protocol

Model predictive control(MPC)is an optimal control based on moving horizon.Since it was originally proposed in 1950 s,with the development over half a century,MPC has been widely used in many fields such as chemical industry,aerospace and medical treatment.On the other hand,with the rapid development of wireless technology and computer science,wireless network has penetrated into social life and gradually replaced the traditional “point-to-point” connection.Therefore,it is of great theoretical and practical significance to study the predictive control of networked systems.However,due to the limitation of network bandwidth,the introduction of network will lead to data collision and network congestion.In order to solve these network induction problems and improve network utilization,it is an effective method to use reasonable scheduling rules during data transmission.However,the introduction of the communication protocol will cause the change of the data update rules,and the data transmission quantity will also be reduced.Hence,some new challenges will be put forward for ensuring the recursive feasibility of the traditional predictive control algorithm and the control performance of the networked system.This thesis studies the robust model predictive control(RMPC)problem for a class of networked control systems under the influence of communication protocols.In order to solve the problems of data transmission,update rule changes and data transmission reduction caused by the influence of common communication protocols such as Round-Robin(RR)protocol and Try-Once-Discard(TOD)protocol,we apply some methodologies such as average dwell time,cone complementary linearization(CCL),stochastic analysis and inequality analysis to design the predictive controller related to communication protocol.In addition,this thesis investigates the network security issues.The predictive controller is designed to ensure the security of networked control systems under the network attack.The main content of this dissertation comprises the following aspects:The RMPC problem under the influence of communication protocols is investigated for the discrete-time linear uncertain system.A set of controllers in the framework of dynamic output-feedback RMPC are designed in terms of the token-dependent Lyapunov-like approach.Some sufficient conditions with less conservatism are obtained by solving an auxiliary optimization problem.An studied for systems with multi-rate scheme and polyhedral uncertainties under RR protocol.By utilizing the lifting technique,the considered multi-rate systems are transformed into the augmentation system with a uniform sampling rate.By employing the inequality analysis technique and RR-dependent Lyapunov-like approach,some sufficient conditions are presented to ensure the system control performance.The controller parameters are obtained by solving the online optimization problem so as to guarantee the proposed algorithm's feasibility and the multi-rate systems' stability.The RMPC problem for polytopic uncertain systems with state saturation nonlinearities is investigated.In order to improve the network utilization and prevent the data collision,the RR protocol is utilized to orchestrate the data transmission order from the controller nodes to the actuator.In terms of the token-dependent Lyapunov-like approach,the controller design scheme is presented to satisfy the robustness of the system.Next,the resilient RMPC strategy is provided for cyber-physical systems(CPSs)with polytopic uncertainties and state saturation nonlinearities under the TOD scheduling.A set of resilient RMPC controllers are designed to guarantee the resilience,robustness and exponential stability.The resilient robust distributed MPC(DMPC)problem is investigated for polytopic uncertain systems under the RR protocol.By decomposing the global system into several subsystems,the computation complexity is successfully reduced.With respect to the existence of various couplings among subsystems,a novel distributed performance index is presented,where information of neighbours at the last iteration is taken into consideration.By using the token-dependent Lyapunov-like approach and inequality analysis technique,some sufficient conditions are proposed.A set of resilient distributed controllers are obtained by solving the online optimization problem and the feasibility of the algorithm and the stability of the distributed systems are ensured.The security-based resilient RMPC problem is studied for a class of discrete-time polytopic uncertain systems with deception attacks and the RR protocol.Both the influence of the adopted RR protocol and deception attacks are seriously considered into the sufficient conditions.A set of controllers are obtained to ensure the resilience and security of the system by solving an auxiliary optimization problem.