Industrial Automatic Control Systems and Controllers Annotation << Back Statistical Identifi cation. Observability and Identifi ability of Dynamical Systems, Algorithmic Feasibility of Adaptive Filtering V.I. Kuznetsov The questions that determine the possibility of the practical applicability of adaptive nonlinear fi ltering under the joint estimation of states and parameters identifi cation of nonlinear non-stationary stochastic dynamical systems close on the feasibility conditions of observability and identifi ability, as well as the algorithmic feasibility of adaptive fi ltering. The possibility of using the local-rank test observability for nonlinear systems with time-varying parameters at each time point of receipt of measurement data based on the infi nitesimal transformations of the phase space of a nonlinear dynamical system in the linear space of differential 1-forms. Substantiated conditions algorithmic feasibility of adaptive nonlinear fi ltering, including the operations of matrix inversion when determining the gain matrix of the nonlinear fi lter and a posteriori estimation of the covariance matrix perturbations in the equations of the model states. Keywords: identifi ability; model; observability; estimate; parameter; condition; fi ltering. Contacts: E-mail: vi_kuznetsov@bk.ru Pp. 12-19.