| In the recent implementation of control systems,digital communication networks are widely used to transmit information due to their lower maintenance cost and more flexible usage and formulated a class of networked control systems.Due to the limitation of feedback network resources,the feedback information can not be available to the controller perfectly,which leads to a decrease in the performance of networked control systems.This dissertation aims to optimize the performance of linear time-invariant system,whose feedback information is transmitted through a rate-constrained communication network.We utilize model-based event-triggered control strategies which can extract the information contained in the sampling time instants to save communication resources and optimize system performance.Compared with the time-driven strategies,our method can guarantee better performance with the same communication resources.The main contents and contributions of this dissertation can be summarized as the following three aspects:Firstly,we consider the problem of exponential practical stabilization of a continuous-time linear time-invariant system in the presence of bounded network delay and process noise.We utilize a model-based periodic event-triggered control strategy which can extract the information contained in the sampling time instants of each dimension to save quantization bit.Sufficient estimation error conditions to guarantee the prescribed rate of convergence are discussed.Moreover,appropriate event-triggered control strategies and corresponding sufficient bit rate conditions that ensure the desired performance are provided in this dissertation.Compared with the time-driven strategies,our method can break the the conventional lower bound of the bit rate condition.Secondly,we aim to optimize the performance of a scalar continuous-time linear time-invariant system,whose feedback information is transmitted through a rateconstrained communication network with random network delay.Due to the limitation of feedback network resources,the feedback information can not be available to the controller precisely and continuously.Moreover,the random transmission delay will reduce the amount of feedback information and degrade the performance of the concerned system.To resolve the above issues,we propose an event-triggered sampling strategy that can utilize the transmission resources more efficiently.Compared with the optimal time-driven sampling strategies and the optimal Lebesgue sampling strategies,our event-triggered sampling strategies can guarantee better performance under the same transmission constraints.Thirdly,we consider the optimization of the performance of the Markovian jump system.The jump of the system models of the Markov jump system brings challenges to the synchronization of quantization and state estimation.We ensure the synchronization of the system state estimation through transmitting the historical information of the system mode.The system performance is optimized by appropriately chooseing the number of quantization bits for different system model.Finally,the explicit expressions among event trigger functions,the number of quantization bit,transmission bit rates and system performance are given. |
