Study Of Adaptive Distributed Fusion Estimation With Non-gaussian Noise

With the rapid development of multi-sensor data fusion technology in information technology,artificial intelligence and control communication,distributed fusion estimation methods have been widely studied and applied due to their advantages of strong robustness,low computational complexity and high fault tolerance.Gaussian noise is usually assumed in the existing distributed fusion estimation,but due to lack of data and information,it is difficult to obtain a Statistical properties of noise or the noise is non-Gaussian in nature.Therefore,this thesis studies the distributed fusion estimation under Unknown-but-bounded(UBB)non-Gaussian noise interference as follows.(1)In the process of distributed fusion estimation of multi-sensor nonlinear systems,when using the information of each sensor to estimate the local state,considering the unknown parameters of the system model,this thesis proposes an adaptive set-membership filter(ASMF).The algorithm firstly uses the Taylor series expansion method to linearize the approximate calculation of the nonlinear part,then integrates the linearization error into the UBB noise,and then uses the set membership identification method to estimate the unknown parameters in the system model online.The Set-membership filter(SMF)algorithm filters the UBB noise to obtain the state estimate,completes the time update and measurement update,so that it can effectively estimate the state and learn the unknown parameters of the system.The simulation results show that,compared with the SMF algorithm,the ASMF algorithm proposed can estimate the system states better.(2)In order to solve the problem that the existing distributed data fusion methods for multi-sensor nonlinear systems do not consider the unknown parameters of the system model,a distributed fusion estimation method based on ASMF is proposed in this thesis.The method first uses the ASMF algorithm to design the local filter of each single sensor,and then completes the design of the global filter based on the Minkowski sum method in the fusion center.The simulation results show that,compared with the traditional distributed fusion estimation method based on the SMF algorithm,the adaptive distributed fusion estimation method based on the ASMF algorithm proposed in this thesis can better estimate the target trajectory.(3)For multi-sensor nonlinear systems with blind sensors,a novel decentralized adaptive distributed fusion estimation method is proposed in this thesis.This method designs a fusion structure in which the sensor preferentially transmits information with its own adjacent sensors so that the fusion center is not required.In this structure,the ASMF algorithm is used to design a local filter in the sensor itself,and the local estimation does not need to enter the fusion center,but is sent to the sensor’s own fusion processor,and the design of the global filter is completed in its own fusion processor.The simulation results show that the decentralized distributed fusion method ignores the existence of the fusion center without reducing the centralized algorithm,so that the blind sensor can receive the information of the adjacent sensors with the operation of the algorithm,which improves the Robustness of adaptive distributed fusion estimation.