ADAPTIVE FILTERING METHOD Russian patent published in 2021 - IPC H03H21/00 G06E3/00 

FIELD: adaptive systems.

SUBSTANCE: invention relates to the field of adaptive systems and can be used for adaptive filtering of stochastic signals and state parameters of stochastic systems. According to the invention, Kalman filtering of the observed vector of the system state is carried out, and at the time of receipt of accurate measurements, the difference between the vector of accurate measurements and the product of the transition matrix of the system state by the evaluation vector at the previous time is formed; the product of the inverse matrix is formed from the product of the extrapolated covariance matrix on the transposed measurement matrix at the previous time by the difference between the vector of exact measurements and the product of the transition matrix of the system state on the evaluation vector at the previous time; a diagonal matrix is formed, the elements of which are defined as the inverse of the corresponding values of the elements of the resulting product (vector); the difference between the Kalman residual vector and the product of the measurement matrix by the difference between the vector of exact measurements and the product of the transition matrix of the system state by the estimation vector at the previous time is determined; an estimate of the elements of the measurement interference dispersion matrix is formed from accurate measurements; the elements of the measurement interference dispersion matrix calculated a priori for the current time are replaced by estimates of the corresponding elements of the dispersion matrix obtained from exact measurements; Kalman filtering of the observed state vector of the system is carried out using the current discrete noisy measurements with a newly formed dispersion matrix of measurement interference until the next accurate measurements are received, after which the adaptive filtering procedure is repeated.

EFFECT: ensuring the stability and increasing the accuracy of Kalman filtering by adaptive determination of the components of the dispersion matrix of measurement interference in the process of current estimation of stochastic signals and state parameters of stochastic systems based on accurate measurements received at irregular (or random) moments of time.

1 cl