Research On Model Predictive Control Strategy Based On Event-Triggered Mechanism For Electromechanical Systems

Model predictive control(MPC)is an advanced control strategy based on optimization theory.Because it can handle the input/output coupling problem of multivariable systems and explicitly consider the physical constraints of system variables,MPC has been widely used in electromechanical system control.However,the traditional periodic MPC method needs to solve the optimization problem online at each sampling time,which will increase the consumption of computing resources and the frequency of data transmission,and it will lead to difficulty in ensuring the control performance of networked intelligent electromechanical systems with limited computing resources and communication bandwidth.At present,the research on event-triggered MPC for electromechanical systems is not yet mature,and there are still lots of rather important problems that need to be solved.For instance,how to design robust state sets and robust terminal controllers to make the system meet state and input constraints;how to design a reasonable triggering condition to ensure the feasibility and robust stability of systems subject to random disturbances and state and input constraints;and for practical electromechanical systems,such as mobile robot systems,how to combine the actual dynamic model and design parameters to improve the availability of event-triggered MPC further,and so on.Therefore,the research of the MPC method based on an event-triggering mechanism has important theoretical significance and application value.This dissertation designs event-triggered MPC methods for linear systems and nonlinear systems respectively,and a novel self-triggered MPC method is proposed for mobile robot systems,aiming to ensure efficient control performance while saving online computing and data transmission resources of the system.The main research content of this dissertation mainly includes the following aspects:(1)For linear time-invariant systems subject to additive disturbances and state and input constraints,an event-triggered MPC algorithm is designed,in which the eventtriggering condition is based on the differential of errors between the actual state and the optimal predicted state.Compared with the common triggering condition based on singletime state error information,the triggering condition based on the differential of continuous-time state errors could further reduce the number of solving optimization problems.Secondly,the iteration feasibility of the algorithm and the stability of the closedloop system are strictly proved in theory,and the Zeno behavior of the system is strictly avoided.Finally,combined with the linear satellite control system,the effectiveness of the scheme is verified through simulation and comparison,which shows that the proposed method can significantly reduce the frequency of solving the optimization and information transmission on the premise of guaranteeing the expected control performance.(2)For nonlinear systems subject to state input constraints and bounded additive disturbances,an event-triggering mechanism based on differential information of system state error is designed to reduce the frequency of the controller to solve the optimization problem,and an event-triggered MPC framework is constructed based on dual-mode control,in which a time-varying robust state constraint is introduced to deal with the additive disturbances so that the predictive control optimization problem has a larger initial feasible region.Subsequently,it is proved that the event-triggered system could avoid Zeno behavior,and the feasibility of the algorithm and the stability of the closed-loop system are proved based on Lyapunov stability theory.Finally,combined with the nonlinear vehicle damping system and continuous stirred-tank reactor system,the simulation results show that this method can significantly reduce the consumption of online computing resources and communication resources while ensuring the desired control performance.(3)Aiming at the conservatism problem caused by an absolute triggering threshold,an event-triggered MPC algorithm with an adaptive mixed threshold is proposed,which further optimizes the control performance by designing a dynamic relative threshold changing with the system state in real-time,to effectively reduce unnecessary sampling and updating of the controller.On this basis,a dual-mode event-triggered MPC framework is designed based on state feedback control,and sufficient conditions are given to ensure the feasibility of the algorithm,the input-to-state practical stability of the closed-loop system,and the absence of Zeno behavior.Finally,the simulation results of the nonlinear building vibration damper system show that the algorithm can save computing resources by reducing the frequency of solving optimization problems,and can increase the maximum allowable disturbance of the system.(4)In order to solve the problem of computing resources and communication constraints in mobile robot systems with constraints and additive disturbances,a selftriggering mechanism is proposed.Compared with the event-triggering mechanism,it can save the cost of state monitoring and thus reduce resource consumption.Subsequently,a self-triggered MPC algorithm is designed based on this triggering mechanism.Through theoretical analysis and research,the influence mechanism of different parameter configurations on control performance and controller triggering condition is revealed,and sufficient conditions to ensure the recursive feasibility of the algorithm are derived.In addition,by ensuring that the time interval for the optimal state trajectory to enter the terminal region is strictly reduced,it is proved that the state trajectory will enter the robust terminal set within the specified time,and the self-triggering condition is obtained based on the stability analysis.The simulation and comparison show that this method can significantly reduce the online calculation and information transmission load of the mobile robot controller without affecting the steady-state control performance.