Innovation approach based measurement error self-correction in dynamic systems

Abstract

Abstract In this work on the basis of the statistics of the relation of selection and theoretical variances, the simultaneous operative testing of mathematical expectation and variance of errors in the one-dimensional measurements are investigated. A new structure and algorithmic provision of continuously operated measurement system with the error self-correction are presented. Mentioned measurement system allows to detect the changes in the statistical characteristics of errors and to carry out their correction on the basis of statistical analysis of Kalman filter innovation sequence in real time. The proposed measurement systems structure with the error self-correction is simpler, as it does not require a model for the errors as that proposed in the existing literature.