Cyclic estimation in primary filtering of sensor signals

The paper is concerned with Kalman estimation of a scalar quantity which is measured together with additive, practically white, noise. The true value of the quantity to be monitored in every short estimation cycle is assumed deterministic and is approximated either as a finite Fourier sum or as a polynomial. For the case of a cyclic Kalman — Bucy filter, KBF, with continuous time inside the cycle simple expresions describe variances of estimation errors at the end of every cycle. For a discrete-time cyclic KBF the necessary computing performance is determined. The proposed filtering maybe implemented in an in-built multiprocessor computer, forecasts variations of the quantity, and detects failures of the sensor and the process.