Podladchikova T. The estimation and prediction of the different-type dynamical processes in the conditions of uncertainty

The dissertation is devoted to the development of system approach to the estimation and prediction of the different-type dynamical processes in the conditions of uncertainty. Its structure provides for application sequence of the developed methods of the long- and short-term prediction adapted to the concrete conditions. The identification methods of the prior unknown noise statistics for nonstationary linear and nonlinear models in the state space and adaptive forecasting algorithms based on these methods are developed. The divergence in the mean-square of the proposed identification methods of the linear non-stationary processes noise statistics is proved. The definition methodology of the potential achievable accuracy of the quasi-optimal smoothing of the indefinite quantities is proposed that makes it possible to estimate the estimation procedure effectiveness. The proposed approach was applied to the problem solution of the estimation and prediction of the different-type dynamical processes namelythe moving object trajectory, the dynamics of the mean arterial pressure and the solar activity dynamics of the 11-years sunspot cycle.