Research On Stability Of Networked Control Systems With Complex Feedback Communication Based On Event-triggering

Since the beginning of this century,networked control systems and event-triggered control strategies have caught much attention and achieved rapid development.In recent years,on the basis of networked control system,information physical fusion system and cloud control system have been developed.Although in practice,the composition of each networked control system may be slightly different,but its common essence is based on the general shared digital network for feedback information transmission.This thesis mainly investigates the control strategy based on event-triggering,which can stabilize the continuous time high-dimensional linear system under complex network environment,and reduce the bit rate used in communication as much as possible.It is noted that the disturbance to the stability of the system under consideration is not only simple process noise,but also more problems that will be encountered in the actual network,such as network delay and feedback packet loss.Due to the coupling of multiple interferences,the stability analysis of the high-dimensional system considered in this paper is more difficult than that of the traditional system only considering single interference,especially in the case of continuous packet loss.Based on the existing research work,this thesis mainly investigates the stability of networked high-dimensional control system in complex network environment.The main contributions of this thesis are presented below.First of all,this thesis proposed a model-based event-triggered quantized feedback control scheme to ensure the(Input-to-state Practically Stability,ISPS)of continuous-time networked control systems with model uncertainty and bounded noise based on event-triggering as the basis of the whole thesis.And the information generated at the sensor is quantized and transmitted to the controller through a digital communication channel,which suffers from the network-induced time delay and clock offset.In order to extend the inter-triggering time of the system,this thesis proposed a model-based event-triggered control strategy and update the control input dynamically with a model of the system plant between two triggering instants.Considering the physical background of the networked control system,a static uniform quantizer is used to encode the data transmitted each time.The information generated at the sensor is quantized and transmitted to the controller through a digital communication channel,which suffers from the bounded network-induced time delay.To make sure two states as close as possible after updating,the information in the "time-stamp" based on network packets is adopted.In this case,we transmit "time stamp" to get the time information of the triggering time,and design an updating method to compensate the model state of the sensor and the controller,so that the two states can stay the same as much as possible after the update.In this thesis we considered the clock offset if the local clock in each node in the real network,such clock offset may cause long-term inconsistency between the state estimates of the sensor and the controller.Furthermore,this paper designed a triggering strategy and updating methods to ensure the input-to-state practically stability of-the network control system under these disturbances.At the same time,based on the above analysis system,there is a non-zero lower bound between two consecutive triggering instant,so the system can avoid the "Zeno-behavior" which indicates infinite triggering in a limited time.And the simulation experiment is designed to verify the effectiveness of this strategy.,Secondly,we proposed a model-based periodic event-triggering control strategy to solve the influence of feedback packet loss and further reduce the average communication frequency of the system,which is built upon the basis of the aforementioned feedback control scheme.In order to ensure the(Input-to-state Stability,ISS)of the system at a finite bit rate,a dynamic uniform quantizer is adapted in this part.In order to use this dynamic uniform quantizer in the system,the triggering condition based on model state is proposed in here.In addition,the triggering function coefficients are modified for adapting the(Independent and Identically Distributed,i.i.d.)feedback packet dropout.Besides,to make sure the two model states equal after updating under i.i.d.feedback dropouts,the(Acknowledgement,ACK)is adopted,included in the control signal,to inform the senor whether the packet is drop or not and modify the coefficients of the event-triggered threshold.With the ACK information,the controller can get the same information about packet loss as the sensor,and make corresponding compensation for the update accordingly.In addition,the finite bit rate which can ensure the system input-state stable is calculated,and the effectiveness of the proposed strategy is verified by simulation experiments.Finally,with the help of periodic event-triggering mechanism,we present a new strategy without using the ACK to ensure the stability of networked control system under i.i.d.feedback dropout.Most of the previous work used the method of passing ACK in the control signal to stabilize the system with packet loss.Acknowledgment signals can inform the estimator the status of control packet losses,which allows the sensor to store the information for compensation in the next transmission.However,the design of ACK is quite complex,so this we proposed the periodic event-triggering strategy and reasonable model updating mode to ensure the input-to-state practically stability of continuous-time scalar linear system with discontinuous packet loss and bounded network delay.The simulation results ensure the effectiveness of the strategy.