Variable Order Filters
Marvin Blum · Defense Technical Information Center (DTIC) · 1971
The theory of variable order filters was derived from the author's previous studies on the application of polynomial splines and generalized splines to statistical filter theory. The current algorithms for the variable order filters are applications of and extensions of the spline function concepts. This class of filters have application to smoothing and prediction of sampled data systems for the following classes of state estimation problems: Estimation of the state vector of a linear dynamic system with rapidly changing trajectories as, for example, a maneuvering vehicle; High precision estimation of the state vector based on long observational intervals where the estimation accuracy may be limited by modelling errors; and, Estimation of the state vector when certain components of the state vector are subject to intermittent discontinuous changes as, for example, staging rockets.