scientific article; zbMATH DE number 3887576

From MaRDI portal
Publication:3222085





[For the entire collection see Zbl 0539.00031.] To understand why and how we introduce the concept of projective filtering as a way to reach feasible algorithms to solve the so called, nonlinear filtering problem, we shall recall the position of the problem as well as the theoritical results about its solutions. We also record that direct use of this result leads generally to infinite dimension algorithms where truncation and approximations are not acceptable because the goodness of such estimators depends on time. Volterra series introduction leads to more acceptable algorithms in some particular situations. Taking advantage of the cascade polynomial structure indicated by the use of Volterra series we try to develop the projective approach. In this approach, the conditional expectation is replaced by the projection of the present state on a space spanned by the polynomial combinations of observations.











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3222085)