A.V. Korennoi - Dr. Sc. (Eng.), Professor, Military Educational and Scientific Center «Zhukovsky-Gagarin Air Force Academy» (Voronezh). E-mail: korennoj@mail.ru A.M. Mezhuev - Ph. D. (Eng.), Associate Professor, Head of Department, Military Educational and Scientific Center «Zhukovsky-Gagarin Air Force Academy» (Voronezh). E-mail: multitenzor@mail.ru

The operation is dedicated to the decision of the task of deriving Bayes of an estimation of stochastic process trajectory on the basis of methods of an optimum filtering of random fields in Gauss approximation. At the decision of the task of an optimum filtering of stochastic processes there is a necessity of deriving of the most precise estimation of a desired signal for any point of a temporal segment of supervision of his trajectory. Till now given task was decided with usage of different methods of interpolation. However their application for deriving potentially of accessible value of error variance of a filtering guesses magnifying of time of supervision and thickening of the circuit of processing. In a case, when the time of supervision is limited and is commensurable with an interval of correlation the known methods do not allow to receive potentially attainable accuracy of an estimation on all a temporal segment of supervision of stochastic process trajectory. In this connection in the given operation it is offered to shape Bayes of an estimation not of separate values of stochastic process se-quentially in time, and all his trajectory on an interval of supervision. Thus the estimations obtained on the basis of the given method, have potential accuracy, at the expense of usage of a posteriori correlation connections between all values of evaluated stochastic process on an interval of supervision. On a concrete example of reception of a mixture Gauss of the Markov process and white Gauss of noise the application of a tendered method is reviewed. The introduced results testifies that the tendered Bayes of an estimations of trajectory has potentially possible accuracy and does not demand padding supervision time. It is his doubtless advantage above known algorithms of a filtering and in-terpolation, and also can be utilised in different data reduction systems, where time of supervision is commensurable with an interval of correlation of evaluated parameters.

 

1. Sosulin JU.G. Teorija obnaruzhenija i ocenivanija stokhasticheskikh signalov. M.: Sov. radio. 1978. 320 s.
2. Kulman N.K.,KHametov V.M. Optimalnaja filtracija v sluchae kosvennogo nabljudenija diffuzionnogo processa s zapazdyvajushhim argumentom // Problemy peredachi informacii. 1978. T. XIV. № 3. S. 55−64.
3. Tikhonov V.I.,KHarisov V.N. Statisticheskijj analiz i sintez radiotekhnicheskikh ustrojjstv i sistem. M.: Radio i svjaz. 1991. 608 s.
4. Korennojj A.V., Ershov L.A. Vosstanovlenie nepodvizhnykh izobrazhenijj kak zadacha prostranstvennojj filtracii staticheskikh sluchajjnykh polejj // Radiotekhnika. 1996. № 7. S. 74−77.
5. Obnaruzhenie, raspoznavanie i opredelenie parametrov obrazov obektov. Metody i algoritmy. (monografija) / Pod red. A.V. Korennogo. M.: Radiotekhnika. 2012. 112 s.