Suboptimal discrete filters for stochastic systems with different types of observations (Q1129498)

The authors consider a discrete stochastic system whose state vector \(x_k\in R^n\) is determined by a linear difference equation \[ x_k=F_{k-1} x_{k-1}+G_{k-l} w_{k-1}, \] where \(F_k\) and \(G_k\) are some matrices, \(w_k\in R^n\) is a Gaussian random vector, \(w_k\sim N(O,Q_k)\). It is supposed that the observation vector \(y_k\in R^m\) is composed of \(N\) different types of observation subvectors. In this connection the linear Kalman filter and the extended Kalman filter are replaced by local filters which are unconnected to each other and allow parallel processing of the observations. The decomposition of the multidimensional observation vector leads to the solution of off-line and on-line computational requirements. A numerical example illustrates the efficiency and high accuracy of the proposed suboptimal filters.