Distributed Fusion Estimator With Time-dependent Observational Noise Uncertain Systems

Time-correlated noises are widely existed in navigation systems and tracking systems.Based on projection theory,measurement differencing and optimal weight-ed fusion algorithms in the linear minimum variance sense,the distributed fusion es-timation problem for multi-sensor stochastic uncertain systems with time-correlated additive measurement noises is studied.The main contents are as follows:For stochastic time-varying systems with AR(1)time-correlated additive mea-surement noises and multiple white multiplicative noises in measurement equations,firstly,the continuous correlated additive measurement noises are transformed into the one-step auto-correlated additive measurement noises by using the method of measurement differencing,and then the linear optimal estimators including filter,predictor and smoother of local single sensor subsystem are designed.Compared with the existing filtering algorithm,the complexity of the algorithm is reduced.The cross covariance matrices between any two local estimation errors are derived,then the distributed fusion estimators in the linear minimum variance sense are given.Further,the multiplicative noise in state equation is considered for the above systems,and the state multiplicative noise and measurement multiplicative noises are correlated.The correlated multiplicative noises are transformed into the one-step auto-correlated and cross-correlated additive noises by introducing a new process noise.For the single sensor subsystem,the local linear optimal filter,predictor and smoother are proposed.The local state filter depends on the recursive filter of the newly introduced p.rocess noise.Furthermore,for the multi-sensor systems,the distributed fusion optimal estimators are given.For multi-sensor uncertain stochastic systems with ARMA(nb,nd)time-correlated additive measurement noises,the original systems are transformed into the systems with state delay and finite-step correlated additive measurement noises by using the method of measurement differencing.The filter,predictor and smoother of the local single sensor subsystem are proposed by introducing the estimators of the products of multiplicative noises and system state.Then,the distributed fusion estimators are given by using the suboptimal weighted fusion algorithms in the linear minimum variance sense and the covariance intersection(CI)fusion algorithm.