Learning Noise Covariances from Prior Parameters and Recursively Collected Data

This paper addresses the recursive noise covariance identification problem by integrating prior parameter information and recursively collected data. Leveraging advanced variational Bayesian inference techniques, we propose a novel prior parameters-based recursive noise covariance identification framework. We then conduct a comprehensive analysis of the identification results, focusing on the effect of recursive collected data and the prior model on the identification performance. Finally, the effectiveness of the proposed method is demonstrated through a simulation study on a three-order system, showcasing its practical applicability and robustness.