Conditionally minimax algorithm for nonlinear system state estimation

In this note, a method of conditionally minimax nonlinear filtering (CMNF) of processes in nonlinear stochastic discrete-time controlled systems is proposed. The CMNF is derived by means of local nonparametric optimization of the filtering process given the class of admissible filters. Sufficient conditions for the existence of the CMNF are considered, and the properties of CMNF estimates are investigated. Results of the CMNF application to control and identification problems are presented.<
>