Design And Application Of Distributed Unscented Kalman Filtering Algorithm

With the wide application of Kalman filtering technology in various industries and the improvement of Kalman filtering requirements in various application scenarios,the estimation accuracy of traditional Kalman filtering algorithms can no longer fully meet the needs of modern society.In some scenarios where the filtering accuracy is not high,the common centralized Kalman filtering algorithm has better applications and performance.However,when the filtering accuracy is high,the centralized Kalman filtering algorithm cannot fully meet the requirements.In response to this problem,this article aims to improve the filtering accuracy of the unscented Kalman filter algorithm.Based on some existing centralized unscented Kalman filter algorithms,the weighted average consensus idea and the finite time consensus idea are respectively introduced to design a new distributed unscented Kalman filtering algorithm.Among them,the distributed unscented Kalman filter algorithm that introduces the weighted average consensus idea includes two types: the unscented Kalman filter algorithm based on the non-square root form of the least sigma points and the unscented Kalman filter algorithm based on the square root form of the least sigma points.By introducing the weighted average consensus idea,the common 2n+1 Sigma sampling points can be reduced to n+1,and the filtering accuracy can be improved while reducing the amount of sampled data.For the distributed unscented Kalman filtering algorithm that introduces the finite-time consistent idea in the multi-agent system,this paper designs two different filtering algorithms for the sensor network with directed communication topology and the sensor network with undirected communication topology.For a strongly connected sensor network,first use the iterative values stored by each node of the sensor network to construct a difference vector,and then construct each difference vector into a Hankel matrix.Combine the core of the Hankel matrix and the idea of consistent ratios to design distributed traceless Kalman filter algorithm;for undirected communication topology sensor network,the distributed unscented Kalman filter algorithm can be obtained by using the core of the Hankel matrix.For the filtering algorithms in the two cases of directed strong connectivity and undirected connectivity,this dissertation has carried out stability analysis.Finally,through Matlab simulation and four-rotor UAV two target trajectory tracking scene experiments to verify the correctness of the algorithm designed in this paper.