Search academic texts

Information × Registration Number 0201U001761, 0197U014521 , R & D reports Title To develop the theory and methods of parametrical ellipsoidal estimation the state and parameters of dynamic systems (method of fuzzy ellipsoids) popup.stage_title Head Bakan G. M., Registration Date 19-02-2001 Organization Space Research Institute popup.description2 The main issue in synthesis of ellipsoidal state and parameter estimation algorithms for dynamic control objects (CO) is method of covering ellipsoids construction that contain intersection of any initial (a priory) ellipsoid with set of special structure determined by measured information features (sets, defined by measuring results). Existing methods allow to build covering ellipsoids (for the case of intersection with sphere) that are optimal according to one or another criteria (volume, matrix track). On the other hand to construct robust estimation algorithms and to provide their convergence it is necessary to have sets of covering ellipsoids that depends on different parameters. In the project on the background of universal methodological approach based on basis functions and mathematical programming methods it is obtained techniques of covering ellipsoids sets construction that explicitly depend on vector or matrix parameters. Suggested idea and technique for covering ellipsoids sets constructi on were used to develop series of new state estimation (state observer and dynamic filter) and parameter estimation algorithms for discrete and continuous CO. Ellipsoidal parameter and state estimation algorithms are developed for discrete and continuos non-stationary multidimensional CO by the results of output measurements. It was assumed that measurement results are disturbed by additive noise of specified magnitude. Reduced state observer and state observer algorithms are developed for multidimensional discrete CO and continuos CO being under non-controlled bounded disturbance of known magnitude. Theorems on convergence of classes of robust state observers are proved for discrete and continuous dynamic CO systems. Problems of convergence conditions determining are reduced to the problems of asymptotic stability of some variables of trivial solution of nonlinear matrix subtract (differential) equation. To solve the last problem extended formalism of Lyapunov-Rumiantsev functions apparatus is used. Comput e r modeling is utilized for suggested algorithms in closed loops of several attitude control systems of spacecraft and prove their effectiveness and reliability. Product Description popup.authors popup.nrat_date 2020-04-02 Close