State Estimation Fusion For Event-Triggered Networked Systems And Its Applications

In recent years,with the increasing development of sensor technology and net-work communication,the trend of control system technology towards network,dis-tribution,intelligence and integration is developing rapidly.Networked systems have emerged as the times require to meet the urgent needs of large-scale industrial production and control process,such as smart grid,cyber physical system,intelli-gent traffic monitoring,unmanned operational command system and so on.How-ever,network-induced problems which include the bandwidth constraint,random packet dropout,transmission delay inherent in the network,bring new challenges to the analysis and design of the networked system,even result in system instability in the worst case.Therefore,there exists a strong requirement for the research on state estimation problem of networked systems with limited bandwidth.Based on the theories and methods of traditional time-triggered sampling systems,the state estimation problems are studied in this dissertation for networked systems with lim-ited communication bandwidth,including the state estimation and fusion problem of event-triggered networked systems,and the state estimation problem of networked systems with quantized information.Moreover,the actuator and sensor fault diag-nosis problem is also investigated for a class of event-triggered dynamic systems.The main contents and contributions of this dissertation are summarized as follows:1.For the state estimation problem of limited-communication networked sys-tems with the existence of structural and parameter uncertainties,an event-triggered stochastic hybrid state estimation algorithm is designed based on the interactive multiple model(IMM)filtering theory.The proposed algorithm can not only effec-tively save network bandwidth and reduce the computational complexity,but also quickly respond to system parameter or structural changes.2.For the state estimation fusion problem of event-triggered networked sys-tems,both sequential and distributed event-triggered fusion algorithms are pro-posed,respectively.The sequential algorithm is a typically centralized fusion struc-ture,while the parallel fusion algorithm with feedback is an optimal distributed fu-sion structure,which has the same fusion estimation performance as the centralized fusion structure.3.For the problem of state estimation fusion for limited-communication net-worked systems with correlated noises,an optimal event-driven distributed fusion algorithm in the sense of linear minimum variance is proposed,which comprehen-sively takes into account correlated noises in practical applications.Besides,it is clear that the linear matrix-weighted fusion criterion in this dissertation is optimal in the sense of linear minimum variance.4.For the problem of state estimation fusion for stochastic event-triggered networked systems,an optimal distributed estimation fusion algorithm is presented based on the maximum posterior probability criterion,which fully takes into ac-count the correlation effect caused by the discretization of continuous systems.The stochastic event-triggered mechanism in this dissertation could strictly maintain the Gaussian characteristic of conditional distribution of the system state.5.For the problem of state estimation for networked systems with bandwidth constraints and network uncertainties,a multi-level quantization-triggered state es-timation algorithm is proposed by considering the coupling influence between quan-tization errors and communication uncertainties.The general multi-level quantiza-tion mechanism designed in this dissertation could deal with the quantization of measurement vectors in parallel with same quantization parameters,and avoid the difficulty of quantization threshold determination in traditional quantization mech-anisms.6.For the problem of actuator and sensor fault diagnosis for a class of event-triggered networked systems,an event-triggered fault diagnosis method is provided based on the interactive multiple model framework.The proposed fault diagnosis method might only establish a candidate fault model for a type of faults,which can effectively reduce the complexity of model set design.The simulation results show that the proposed algorithm has a fast response speed for the fault occurrence,and accurately estimates the fault magnitude of the faulty actuator.