Research On The Information Fusion Estimation Problem Of The System With Unknown Model Parameters

For system models with unknown parameters,direct state estimation is not possible.Because the state estimation of the system model requires the model parameters and the noise variance to be known,the unknown parameters in the system model are identified before the state estimation is performed,and the result of the parameter identification is brought into the system model,and then the system is The model performs state estimation.In this paper,different parameter identification methods are adopted according to different system models with unknown parameters to improve the accuracy of parameter identification.The multi-incident identification theory is to expand the interest rate from single innovation to multiple innovations to improve data utilization.Rate to improve the accuracy of parameter identification.However,the information collected by a single sensor is sometimes inaccurate and incomplete,and cannot truly reflect the characteristics of the system.Therefore,multi-sensor information fusion technology plays an important role in state estimation.Multi-sensor information fusion is to fuse the collected information from multiple sensors under certain criteria,so that the information is more fused and more accurate than the single sensor,and the anti-interference ability is stronger,so that the multi-sensor system is optimal.Convergence estimates.In this paper,the research on different types of multi-sensor systems with unknown parameters mainly includes the following aspects:Firstly,for the canonical control model multi-sensor system model with unknown parameters of state matrix,the formula based on state space model is transformed into ARMA model and the general recursive least squares parameter identification algorithm is given.To identify the accuracy,a multi-innovation recursive least squares algorithm and an improved multi-innovation least squares algorithm are proposed.Based on this,Kalman filtering,prediction and smoothing estimator based on the parameter identification results of the model are given.The formulas of distributed state fusion estimator and centralized observation fusion estimator are given.Secondly,for the observable controlled multi-sensor system model with unknown parameters for the state matrix and the left multiplying matrix of the control input,the recursive augmented least squares algorithm and the iterative minimum two are proposed for estimating the unknown parameters of the system model.The multiplication algorithm is given and the Kalman filtering and prediction state estimator based on this model is given.Based on this,we give a Kalman filter state estimator based on iterative least squares algorithm and a Kalman prediction state estimator based on iterative least squares algorithm.In order to improve the state estimation accuracy,we use the weighted observation fusion algorithm to give the formula of the multi-sensor weighted observation fusion iterative least squares Kalman filter estimator and Kalman prediction estimator algorithm.Finally,the RGIA algorithm for model unknown parameter identification is given for the multi-sensor model of atypical paradigm system with unknown parameters.Based on the parameter identification algorithm of RGIA algorithm in SISO system model and the parameter identification algorithm of RGIA algorithm in state space system model,the state estimation results and the state estimation of RGIA in multi-sensor system are obtained based on the parameter identification results.The effectiveness and correctness of the algorithm are verified by several simulation examples.