Research On Event-triggered State Estimation For Nonlinear Networked Systems

With the rapid development of modern communication technology,the structure of traditional control systems is changing,and networked control systems(NCSs)that combine network with traditional control systems are emerged.Compared with traditional control systems,NCSs have the advantages of higher flexibility and better resource sharing.Meanwhile,there also exist many uncertainties for NCSs such as limited communication bandwidth,network-induced delay,correlated noises,network attacks and so on,which results in the research on state estimation for NCSs more challenging.Moreover,the improvement of sensors performance also makes the research of fusion estimation for multi-sensor networked systems becoming one of the most important research contents.It is worth emphasizing that,due to the complexity of control objects,many control systems usually show different degrees of nonlinearity,which further increases the design difficulty of state estimators.Aiming at the above problems,a series of unscented Kalman filter(UKF)-based nonlinear filtering and fusion estimation algorithms are designed in this paper,and corresponding estimation performance is analyzed.Main research contents include:Firstly,the filtering problem for nonlinear networked systems with event-triggered data transmission and correlated noises is considered.An event-triggered data transmission mechanism based on measurement innovation is introduced to reduce excessive measurements transmitted over a wireless network.Consider that process noise and measurement noise are one-step cross-correlated,an UKF-based filtering algorithm that depends on correlation parameter and trigger threshold is presented.Then sufficient conditions are established to ensure boundedness of the estimation error covariance.Secondly,an event-triggered distributed fusion estimation problem is investigated for a multi-sensor nonlinear networked system with random transmission delays.For each communication channel,an event-triggered scheduling mechanism is introduced to reduce excessive measurement transmission,and a fixed length buffer is used to retrieve partly delayed measurements.Then a distributed fusion estimation algorithm is designed using a sequential covariance intersection fusion strategy,and sufficient conditions are also established to ensure boundedness of fusion estimation error covariance.Finally,a stochastic event-triggered distributed fusion estimation issue is investigated for bandwidth-constrained multi-sensor nonlinear networked system suffering from jamming attacks.For each communication channel,a stochastic Send-on-Delta(SoD)event-triggered transmission scheme is developed to reduce excessive communication between smart sensors and local estimators,and a Stackelberg game framework is established to analyze interactions between a smart jammer and smart sensors.Utilizing a sequential fast covariance intersection fusion technique rule,a distributed fusion estimation algorithm is designed,and convergence conditions on the designed fusion estimation algorithm are further determined.