Research On Distributed Target State Estimation Method Under Non-ideal Conditions

The target state estimation belongs to the research field of target tracking.It is the second step of anti-aircraft fire control systems,which include the target search,track,hit and damage.Therefore,it is necessary for the interception task of anti-aircraft fire control system to track the target timely and effectively.The distributed target state estimation has been more and more concerned and applied for its high efficiency,scalability and invulnerability.However,in the practice,the distributed detection system may have the problem of estimation accuracy degradation,due to the complexity of the working environment,the limitations of the detection equipment or other factors.And finally it is difficult to track the target.According to the network structure and working characteristics of distributed detection system,the problems of distributed target state estimation under the non-ideal conditions of limited energy,incomplete measurement and asynchronous sampling were studied in this thesis.Through a series of improvements and a variety of fusion estimation methods,this thesis aims to improve the performance of distributed target state estimation algorithm.Finally,the results of this thesis can offer the theoretical basis and technical guidance for the improvement of China’s distributed anti-aircraft fire control system.The main contributions of this thesis are summarized as below:(1)Considering the energy limitation in the detection network,the design of multi-frequency fusion estimation algorithm based on event-triggered mechanism and matrix weighting is studied.Firstly,For a class of linear time invariant systems,a multi-rate target state estimation model is constructed with the lifting technology,which can effectively reduce the information interaction frequency between the detection units.Then,considering the relevance between the local estimation of each detection unit,a two-stage matrix weighted fusion algorithm is proposed with the matrix weighted fusion method.For the further reduction of communication burden,an event-triggered mechanism based on local estimation and first-stage fusion estimation is designed.In order to achieve the balance between the fusion accuracy and the information interaction frequency,the event-triggered threshold can be dynamically adjusted by setting the containment probability.Finally,the numerical simulation results show that the proposed two-stage multi-frequency distributed state estimation algorithm based on event-triggered can ensure the estimation accuracy and reduce the burden of the communication system,which is feasible to be applied.(2)Considering the phenomenon of the intermittent observations and asynchronous data interaction in the detection systems,the design of sequential Kalman consensus filtering algorithm and the stability of the algorithm was studied.Firstly,with the unknown crosscovariance between local estimations,a sequential covariance intersection-based Kalman consensus filter with the intermittent observations was proposed.Using this algorithm,each detection unit can obtain the consistent fusion results sequentially,and the fusion estimation accuracy is independent of the fusion order.Then,it is proved theoretically that if the observation arrival probability of the system exceeds a given threshold,the estimation error variance is bounded.Meanwhile,based on some practical assumptions,it is strictly proved that under this condition,the estimation error of the algorithm is exponentially bounded in mean square.Finally,two groups of numerical simulations verify the effectiveness of the algorithm and the correctness of the theoretical results.(3)Considering the problems of intermittent observations and nonlinear filtering in the detection networks,the design of distributed nonlinear consensus filtering algorithm and the stability of the fused results were studied.Firstly,the cubature Kalman filter algorithm with intermittent observations was given.With the covariance intersection method and the consensus filtering,the consensus on information distributed cubature Kalman filtering algorithm with intermittent observations was proposed.Then,the estimation error variance matrix is derived by Taylor expansion,and it has been proven that the error covariance matrix will be bounded if the observation arrival probability exceeds a given threshold.Finally,proposed algorithm is applied to a class of distributed optoelectronic tracking network.The experimental results of two groups of typical moving model targets show that the proposed algorithm can track the nonlinear target,and have higher estimation accuracy and consistent results compared with other fusion algorithms.(4)Considering the problem of the asynchronous sampling in the detection networks,the design of asynchronous matrix weighted fusion algorithm was studied.Firstly,the augmented state space model of asynchronous sampling system is established with the method of observation augmentation.And the local estimation algorithm was given.Then,the method of calculating the cross-covariance between the local estimations was derived.Furthermore,based on the augmented estimation model and the statistical characteristics of local estimation,an optimal sequential fusion estimation algorithm based on matrix weighting is proposed.Finally,the numerical simulation results show that the proposed algorithm can solve the distributed estimation problem of asynchronous sampling system,and it can ensure the estimation accuracy with lower computational complexity.