Search

In this article, the distributed adaptive Kalman filtering is investigated for discrete-time stochastic nonlinear systems with gain perturbation as well as unknown covariance of process noises. For the adopted event-triggered communication scheduling, a distributed Kalman filter with an event timestamp is first constructed to effectively fuse the information from neighbors and itself while guaranteeing the unbiasedness. In light of stochastic analysis, the desired filter gain, achieving the suboptimality of filtering performance, is obtained recursively by solving two optimization issues with the form of Riccati-like difference equations. With the help of the fashionable weighted fusion conception combined with the well-known law of large numbers, a recursive estimation of process noise covariance is derived step by step and consequently suits for online computation. Finally, the effectiveness of the proposed filtering scheme is verified via a “lineland” system model. [ABSTRACT FROM AUTHOR]

The distributed recursive filtering problem is investigated in this paper for discrete time-delayed nonlinear stochastic systems, where the well-known Takagi–Sugeno (T-S) fuzzy model is used to approximate the nonlinearities. According to obtain the system dynamics, a novel structure of distributed filters is developed, where the difference of estimated states from neighboring sensors is exploited to improve the one-step prediction, and the desired estimation is obtained by fusing the estimation under different rules. Attention is focused on the design of a distributed recursive filter such that, in the presence of time-delays and defuzzifying operations, an upper bound of the filtering error covariance is obtained and then minimized by properly designing filter parameters via elaborate mathematical analysis. With the exception of the desired gains with the online recursive form are dependent on the solutions of two Riccati-type difference equations, and the upper bound is further optimized via the introduced parameters. As a final point, a simulation examples is exploited to show the applicability of the developed filtering scheme. [ABSTRACT FROM AUTHOR]

In this paper, the design of networked recursive filter is investigated for a kind of nonlinear stochastic systems subject to missing measurements, fixed time-delay and uniform quantisation under the Round-Robin protocol. In a digital platform, the information coming from multi-sensors could be subject to missing measurements due to the environment effect and then need to be transformed into digital signals via uniform quantizers. In addition, the Round-Robin protocol is adopted to govern the token accessing to channel media so as to both save energy and mitigate the data congestion. A novel extended Kalman-type recursive filter is constructed that firstly combines the Round-Robin protocol, namely, in the form of the augmentation of the plant system dynamics and the protocol-induced periodic measurements. In the simultaneous consideration of system delays, missing measurements, uniform quantisation, as well as the Round-Robin protocol, the purpose for the discussed filtering problem is to obtain a set of filter parameters over a finite-horizon to minimise the upper bound of filtering error covariance as far as possible. Via elaborate mathematical analysis, the desired filter parameter is obtained by virtue of solving two Riccati-type optimisation equations, which are dependent on the latest estimation states. The genetic algorithm has been introduced to optimise the dynamic parameter selections. In addition, it is revealed in theory that the trace of the upper bound of filtering error covariance is non-decreasing as the quantisation level increases. Finally, the effectiveness of the proposed design scheme is inspected by a discretised maneuvering target tracking system. [ABSTRACT FROM AUTHOR]

In this paper, the recursive filtering is investigated for a type of time-delayed stochastic nonlinear systems with event-based communication protocols and missing measurements. A varying condition threshold in event triggering protocols is introduced to better match up with the system dynamic performance. The purpose of this paper is to develop an easy-implemented recursive algorithm with consideration of linearization errors, time-delays, packet losses as well as adopted communication protocols. By carrying out some elaborate mathematical operation, the desired parameters can be derived by means of the solutions to two Riccati-like difference equations. Such parameters can effectively suppress the trace of filtering error covariances and therefore the developed filtering algorithm is suboptimal. Finally, an illustrative example is presented to show the effectiveness of the developed filtering algorithm. [ABSTRACT FROM AUTHOR]