Multi-sensor Optimal Estimation And Fusion Algorithm

Due to the advent of the sensor technology and communication technique,multi-sensor systems have recently attracted considerable attention, especially, thosewith diversified complex using background. State fusion estimation is an importantstudy field in the information fusion theory, mainly dealing with how to estimate thesystem state exactly by multi-sensors. It is usually applied in tracking system and otherexact estimating systems.This thesis considers optimal estimation and fusion algorithm of multi-sensor datafusion, and deeply studied the issues related to which. Some familiar system structureswere introduced briefly, this thesis respectively introduced the optimal fusionalgorithms under the distributed and centralized system structures, and then analyzedand compared the performance of the different fusion algorithms.This thesis systemically studies the multi-sensor with correlated measurementnoise. When the measurement noise covariance is certain matrix that can be transformedto a diagonal matrix by matrix resemble transform, this thesis develops optimalcentralized and distributed fusion estimation. Based on this, a new fusion method thathas equivalence compared with the centralized optimal fusion algorithm is presented, onsome special condition, computational burden can be reduced by the fusion algorithm,so it is useful for some application. Simulation showed the equivalence between theoptimal centralized fusion estimation algorithm and the new fusion algorithm.Aimed at the state fusion problem of nonlinear dynamical system with multi-sensor,this thesis introduces a weighted multi-sensor data fusion algorithm on the condition ofthe minimum mean square error(MMSE), and then the weighted algorithm based onUnscented Kalman Filter(UKF)method is proposed. The simulation results show thatthe proposed weighted UKF state fusion method is outperform to the Extended KalmanFilter(EKF) ones.