Distributed State Estimation And Target Tracking Control Based On Kalman Consensus Filter

With the advantages of flexibility,reliability,scalability,and low cost,multi-sensor networks have promising applications in environmental monitoring,motion estimation,mission search and rescue,industrial processes,and location navigation,and have therefore received a lot of attention from scholars.All these scenarios of the application require multi-sensor networks to provide accurate estimation and tracking of target states.It is also a challenging problem to design a suitable distributed state estimation algorithm when the sensor nodes in the network have different models,i.e.,the estimated states are not in the same state space.Based on this,this thesis firstly investigates the problem of state estimation for multi-sensor networks with distributed structure.Further,a filtering algorithm is designed to solve the distributed state estimation problem in different state spaces by using the mainstream Kalman filtering algorithm theory and consistency strategy.Finally,the target tracking algorithm is designed by estimating the target state through the mobile multi-sensor network.The main work and results of this thesis are as follows:By indirect state vector fusion,the Kalman consistency filtering algorithm is extended and a new collaborative filtering algorithm is proposed for multi-sensor networks where each sensor node has a different model but there is a known affine transformation relationship between the models.In this thesis,the distributed Kalman consistency filtering algorithm is studied based on the distributed sensor fusion approach and the Kalman filtering algorithm.The stability of the model in the absence and presence of noise is analyzed by the Lyapunov function,and the deterministic bounds of the error covariance matrix and the Kalman gain matrix are also derived.The collaborative filtering algorithm designed in this thesis is applied to the practical problem of relative positioning of multiplayer unmanned aircraft system based on a single landmark,and the feasibility and practicality of the filtering algorithm are finally illustrated.For the case that controlled object is a discrete-time model and the target state cannot be obtained directly,the target state is estimated by moving a multi-sensor network and based on the cooperative filtering algorithm designed in this thesis,and a target tracking control algorithm with disturbance term estimation is designed based on the discrete sliding mode control theory.The upper bound of the variation amplitude of the disturbance term is determined for whether the disturbance term contains a Gaussian noise term or not,and the selection method of the controller parameters is given to analyze the stability of the closed-loop system in the above two cases.Finally,the effectiveness of the filtering algorithm as well as the control algorithm designed in this thesis is verified by completing MATLAB simulation experiments.