Distributed state estimation for uncertain linear systems: A regularized least-squares approach

Self Cite

This article investigates the distributed maximum correntropy unscented Kalman filtering problem for nonlinear systems via a sensor network. The system dynamics is subject to state equality constraints and non-Gaussian noise. By utilizing the maximum correntropy criterion to handle non-Gaussian noise, a centralized maximum correntropy constrained unscented Kalman filter is first proposed. Then, two novel distributed maximum correntropy constrained unscented Kalman filters with special features are designed. Specifically, the first one is developed by approximating the centralized filter with each sensor's own and its neighbors' measurements. The other one is designed by fusing state estimates. It is worth mentioning that these two distributed algorithms only need finite steps to fuse information over the sensor network rather than infinite steps to achieve the average consensus. Finally, the validity of the proposed algorithms is demonstrated by simulation experiments, with a detailed comparison.

Section: Problem Statementmentioning

confidence: 99%

Distributed maximum correntropy unscented Kalman filtering with state equality constraints

Duan

et al. 2021

Self Cite

“…[6][7][8][9][10] This increases system redundancy and robustness, and makes parallel computing possible. Consensus-based Kalman filtering, [11][12][13][14][15][16][17] as a popular method to address the problem of distributed state estimation, achieves acceptable estimation performance for all sensor nodes. For instance, He et al 14 propose a suboptimal algorithm based on distributed Kalman filter, and adaptive fusion weights are designed.…”

Section: Introductionmentioning

confidence: 99%

“…For instance, He et al 14 propose a suboptimal algorithm based on distributed Kalman filter, and adaptive fusion weights are designed. Duan et al 15 address the distributed filtering problem for an uncertain system based on a regularized least-squares approach. In References 14 and 15, a global observability condition is presented for linear systems, which relaxes the local observability of each sensor.…”

Section: Introductionmentioning

confidence: 99%

Distributed unscented Kalman filtering for nonlinear systems: A mixed event‐triggered strategy

Sun

Yang

2021

In this article, the problem of distributed state estimation is addressed for nonlinear systems based on the unscented Kalman filter (UKF) framework. The local estimates are obtained from distributed UKFs with stochastic event-triggered schedules. The consensus of estimated states is achieved by employing the covariance intersection fusion method, while a novel event-triggered strategy for the node-to-node communication is developed. Furthermore, the boundedness and convergence of the covariance matrix are analyzed. Compared with the existing works on distributed Kalman filtering, a new nonlinear global observability condition is proposed to relax the constraint that the system is necessarily observable by each sensor, and the relationship between the convergence and the event-triggering parameters is revealed. Finally, simulations demonstrate the results in this article.

“…A hybrid consensus filter by combining two approaches of consensus on measurements and consensus on information was considered 19 . By utilizing a regularized least‐squares approach, a distributed state estimation for uncertain linear systems was proposed 20 . An extended state based distributed Kalman filter and an event‐triggered distributed filter for a class of stochastic uncertain systems over switching sensor networks with quantized communications were proposed 21 …”

Section: Introductionmentioning

confidence: 99%

Distributed Kalman filtering for uncertain dynamic systems with state constraints

Duan

2020

Self Cite

This article addresses the distributed state estimation problem for uncertain time-varying dynamic systems with state constraints over a sensor network. By using a null space method, the distributed state estimation problem for uncertain dynamic systems with state constraints can be cast into a new unconstrained distributed state estimation problem for reduced uncertain dynamic systems. A constrained distributed Kalman filter is proposed, and it is shown that the full state estimates can be recovered at any time and satisfy the constraints. An optimized upper bound of the estimation error covariance of each sensor is obtained, and the corresponding gains are designed. The application conditions of the proposed algorithm are mild, and they can be off-line checked. Furthermore, the computational requirements in this article are also reduced compared with the existing results. Finally, the performance of the proposed filter algorithm is demonstrated through numerical simulations.