K.P. Likhoedenko – Dr. Sc. (Eng.), Professor of the chair of (sub)department «Autonomous information and operating systems» of Bauman Moscow State Technical University
V.K. Khokhlov – Dr. Sc. (Eng.), Professor of the chair of (sub)department «Autonomous information and operating systems» of Bauman Moscow State Technical University
In article problems of synthesis of algorithms of systems of decision-making in autonomous information systems of AIS of the near location (NL) taking into account specifics of their work are considered. In AIS NL informative parameters of signals often are uncentered random variables or processes for which population means a priori aren't known, estimates of population means on one realization are often impossible because of high speed of such systems. Therefore in article the generalized criteria of synthesis of AIS are considered, the private criterion for synthesis of a path of decision-making of AIS NL is reasonable. The algorithm of work and the generalized block diagram of AIS are proved. Taking into account specifics of work of AIS regression statistical characteristics of uncentered parameters of signals and hindrances–coefficients of multiple initial regression are proved, regression algorithms of detection and recognition of signals are given. In regression algorithms aprioristic information on correlation properties uncentered parameters of signals and hindrances is considered through coefficients of multiple initial regression. Problems of choice of informative signs in AIS which are the most effective from the point of view of divisibility of classes of signals and hindrances, and questions of reduction of dimension of vectors of signs are important. In article the technique of a choice of informative signs and reduction of dimension of vectors of signs in relation to the algorithms using initial estimates of casual parameters is offered. The regression algorithms offered in article allow to process non-stationary casual signals at unknown population means of informative parameters. The problems considered in article open new approach to creation of paths of processing of signals of noiseproof perspective AICS.