Event-triggered Predictive Control Analysis For Stochastic Time-delay Systems

With the advance and expansion of communication technology and control theory,networked control systems have gradually replaced the traditional point-to-point transmission structure to achieve resource sharing among sensors,controllers,and actuators through the network and make the system more flexible and reliable.However,due to the involvement of unreliable networks,new problems have also been brought,such as time delay,channel noise interference and so on.These problems may restrain function of the system and destroy stability of the system even more.In addition,in the era of big data,the amount of data and information manifests an explosive growth,so it is very necessary to filter data under limited network resources.Addressing the aforementioned issues in networked control systems,the thesis puts forward two solutions.Firstly,constructing an event-triggered predictive control strategy under Kalman filtering for discrete stochastic systems with control delay,and the Schur stability of the system is obtained.Secondly,raising an event-triggered predictive control strategy with time regularization in continuous stochastic systems with output delay,and the closed-loop system is exponentially stable in the mean-square sense.Finally,we examine the validity of the obtained theoretical results through simulation studies such as satellite control system and single-link manipulator system.The specific works and innovations are as follows:1.For discrete stochastic systems,the thesis puts forward an output-based event-triggering mechanism to filter the actual output values,so that the occupation of network bandwidth resources and control costs can be reduced.For continuous stochastic systems,a time regularization method is added to event-triggering mechanism to prevent the control from updating infinitely in a finite time which can effectively avoid Zeno phenomenon.2.This thesis comes up with a Kalman filtering based on event-triggering.After filtering measurement noise and process noise,the filter can estimate the system state on the basis of the actual output value selected by event-triggering,and its gain matrix updates with the update of observation data,which better improve the utilization of communication resources.3.For discrete stochastic systems that exist d-step control delay,an event-triggered predictive control strategy under Kalman filtering is proposed.The predictive controller predicts the output information of controller at k time by the input information at k-d time.For continuous stochastic systems with output delay,a dynamic prediction mechanism is designed.And combing with event-triggered control,the stability of the system is achieved.