Enhanced Recursive Total Least Squares Method With Subspace Tracking and Noise Covariance Estimation

In this letter, we present a novel recursive approach aimed at refining parameter estimation within the Errors-in-Variables (EIV) framework. Our method integrates the Weighted Total Least Squares (WTLS) technique with a Subspace Tracking algorithm and a dynamic Noise Covariance Adaptation solution, offering improved accuracy and precision in estimating parameters, particularly when errors affect both input and output variables. By merging WTLS and Subspace Tracking methodologies, our model adeptly adapts to variations in system dynamics. Moreover, the integration of a real-time Noise Covariance Adaptation mechanism into our parameter estimation strategy effectively addresses uncertainties stemming from input and output measurement noise. Through different simulations and comparative analyses, we validate the efficacy of our approach, underscoring its potential to significantly advance parameter estimation within EIV models across diverse applications.