Engineering Physics Annotation << Back PECULIARITIES OF APPLICATION OF THE KALMAN FILTER IN THE PROBLEMS OF COMPLEX PROCESSING OF THE INFORMATION A.V. BABICHENKO S.Y. SUСHORUKOV V.K. SHKRED M.I. ORECHOV Provides an overview of the theory of random functions: correlation function and its properties, the model is absolutely random process of white noise, the Markov property of random process-es, the equation of state of a linear dynamical system perturbed by Gauss noise. Is well-known conclusion of the algorithm of the estimation of normally distributed random vectors in Gaussian measurements, statistically related to the measured vector. On the basis of the information given above is given an algorithm of discrete Kalman filter – as the algorithm of estimation of the condition of a dynamical system in the classical approximation. Provides easy-to-practice poskalyarnoy measurement processing algorithm and the conditions for its realization. Lists the main problems of realization of the Kalman filter, associated with the variation of the physical object and mathematical statement of a problem of estimation, the decision of which is a filter. Describes the major well-known ways of ensuring computer sustainability of digital implementations filter: modification of Joseph, modification on the basis of the verkhne – or lower triangular decomposition the covariance matrix, modification on the basis of треугольно-diagonal of the decomposition of the covariance matrix, which use ортогонализацию gram-Schmidt. Are some implemented in practical ways of formation of adaptive robust modifications filter on the basis of harmonization of the estimated and the actual residual covariance. Given the elements of covariance analysis and its application to the problem of synthesis of algorithms of filtering. With sufficient for practical applications in detail considered covariant equations of errors for the General case description of the error models and measurements. Demonstrates the use of these expressions in the synthesis of on-Board algorithms for the integrated processing of navigation systems on the basis of minimax criterion of the quality of the filter under a priori setting the method for mathematical simulation of the typical trajectories. Key words: random vectors and processes, the equation of state, the evaluation of a random vector, Kalman filter, the computational stability, the adaptation of the filter, covariance analysis, minmax quality criteria. Contacts: E-mail: rpkb@rpkb.ru Pp. 35-50.