Research On Two-stage Kalman Filtering Fusion Algorithm For Complex Systems

The traditional Kalman filtering algorithm is mainly used in various systems that do not consider bias.If you insist on using it in a system with bias,the filtering estimation results are not ideal.The augmented state Kalman filter and the two-stage Kalman filter are one of the few algorithms that can be better applied to the system with bias.Among the various extensions and applications of the above two algorithms,the problem of noise correlation is less mentioned.After various engineering applications and theoretical studies,it is shown that the simple assumptions of various noises of the system are no longer enough to meet people's requirements.At the same time,the research on filtering algorithms mostly focuses on linear systems,and linear systems are only a special case.Undoubtedly,various types of nonlinear systems are also facing the effects of bias.Therefore,the study of the filter estimation algorithm for nonlinear systems with deviations has also become meaningful.In response to the above problems,we did the following work:(1)A two-stage Kalman filtering algorithm for noise-related systems is proposed.For a class of bias estimation systems with correlated noises,a two-stage Kalman filter capable of dealing with noise correlation is proposed.By considering the objective facts related to measurement noise in process Engineering,based on the traditional two-stage Kalman filter without correlated noises,a noise decorrelation technique is introduced to establish a system bias estimation method with wider applicability.(2)A two-stage Kalman filter fusion algorithm for noise-correlated multi-sensor systems with bias is established.For the multi-sensor system with dynamic bias and correlated noises,based the two-stage approach,proposing a new fusion method:Firstly,multiple state filters without bias and bias filters are combined in parallel,and then the fusion results are combined to obtain the estimated value of system state.(3)A nonlinear fusion method based on two-stage cubature Kalman filtering for a class of nonlinear multi-sensor systems with bias is proposed.Aiming at the nonlinear system with dynamic bias widely existed in the real world,based on the traditional two-stage cubature Kalman filter,a decentralized filtering technique is introduced to propose a nonlinear filtering estimation for nonlinear systems with dynamic bias.