Adaptive Filtering in Dynamic Systems. Guaranteed Approach

This paper considers the creation of a minimax filtering algorithm in the form of an adaptive three-level algorithm consisting of several filters at each level for dynamic systems (DS) under uncertainty of the initial state caused by disturbances acting on the DS and by interference in information channels during operation. The filters are built based on either stochastic or deterministic information extensions of the process model. Same-level filters and the levels between themselves are covered by feedback, which allows for the adjustment of the a priori specified parameters. Thus, a filterbank is generated and the filters are integrated, which enhances the adaptability of each processing stage. The efficiency of the algorithms is exemplified.