Adaptive filter parameter reconstruction technology for rocket inertial navigation/satellite integrated navigation system

The integration of micro-electro-mechanical systems (MEMS) strapdown inertial navigation systems (SINS) with global navigation satellite systems (GNSS) has emerged as a significant area of research due to its compact size, affordability, and high precision. In the context of guided rocket-borne MEMS-SINS/GNSS integrated navigation systems, the performance of navigation is characterized by the need for high overload, accuracy, and real-time capability. A variety of enhanced algorithms based on Kalman filtering are currently employed as integrated filtering methods, which comprehensively address deviations in the system model to improve navigation performance. The noise characteristics of MEMS inertial guidance devices change dramatically under long-term storage conditions, while the dynamic flight environment of rockets and the high real-time requirements of navigation solving make the design of on-board combined navigation filters challenging. To address this issue, this article introduces the Adaptive Reconfigurable Extended Kalman Filter (AREKF) method. Initially, a precise system state model is developed to reflect the unique characteristics of the rocket flight environment, facilitating rapid convergence of the filtering process. Subsequently, during the rocket alignment process, a real-time reconstruction of filter parameters is implemented to enable adaptive and precise modeling of navigation parameters. This strategy ensures lower computational costs during rocket flight, enhances the accuracy of the navigation system, and produces real-time navigation outputs that exhibit high overload and precision. The results from the Six-Degree (6D) Model simulation and car-mounted experiments demonstrate that, compared to the traditional Extended Kalman Filter (EKF) algorithm and existing improved algorithms, the AREKF method significantly enhances the real-time navigation accuracy of rockets under high overload conditions.

Abstract licence: CC BY

A person counts once a day per paper. We keep no cookie and no address: a reader is a hash that changes every day. Any month under five reads shows no number.