Aerospace Instrument-Making Annotation << Back Neuromorh multimodel algorithm of the vector graviinertial navigation system correction Devyatisil’niy A.S. Chislov K.A. The problem of evaluation of the Earth’s gravitational field strength on the research vehicle trajectory is considered. In the generality of theoretical and mechanical concepts underlying the inertial navigation and gravimetry, the solution of this problem is immersed in the problem of the three-component inertial navigation system (3D-INS) dynamic block correction by the full positional information delivered from satellite navigation system GLONASS. A mathematical model of the correction problem is identified with the inverse problem of the «state-measurement» type. Dynamic inversion of such problem is executed by the artificial neural network (NN). Functioning of the NN associated with known properties that characterize thinking, such as focusing, modularity, habituation. The algorithm of dynamic inversion bases on the multi-model interpretation of the Kalman filter, which causes the customization mechanism of NN’s synaptic coefficients and ensures stability of the inverse problem solution. In order to verify the proposed neural network for correction of the vector gravimetric 3D-INS and illustrate its effectiveness the numerical experiment was carried out, its description and results are given. Key words: inertial navigation system, neural network, neuromorphism, gravimetry, Kalman filter. Contacts: E-mail: kirillche@rambler.ru Pp. 27-33.