Towards Active Excitation-Based Dynamic Inertia Identification in Satellites

TL;DR

This work tackles the problem of identifying time-varying inertia properties of satellites in flight, a task crucial for reliable attitude control and sensitive to how torques are applied. It conducts a broad, simulation-based comparison of eight torque profiles and two estimators—batch Least Squares under a static inertia assumption and an Extended Kalman Filter that tracks inertia as a slow random-walk—across three satellite configurations and several inertia-change scenarios. The study demonstrates that excitation bandwidth and estimator assumptions jointly govern identification accuracy: smooth, spectrally rich excitations favor LS, while dynamic, broadband excitation enhances EKF performance, especially for larger satellites; results are reinforced with open-source code for reproducibility. The findings provide actionable guidelines for in-orbit inertia identification strategies and highlight avenues for future work, including constrained estimation and reinforcement-learning–driven excitation design.

Abstract

This paper presents a comprehensive analysis of how excitation design influences the identification of the inertia properties of rigid nano- and micro-satellites. We simulate nonlinear attitude dynamics with reaction-wheel coupling, actuator limits, and external disturbances, and excite the system using eight torque profiles of varying spectral richness. Two estimators are compared, a batch Least Squares method and an Extended Kalman Filter, across three satellite configurations and time-varying inertia scenarios. Results show that excitation frequency content and estimator assumptions jointly determine estimation accuracy and robustness, offering practical guidance for in-orbit adaptive inertia identification by outlining the conditions under which each method performs best. The code is provided as open-source .

Figures (4)

• Figure 1: Overview of torque excitation profiles used for system identification. The figure shows eight profiles plotted as simultaneous signals: One step, Multi step, Sawtooth, Sine, Multi-sine, Chirp. PRBS, and Sine 3-axis. These profiles have been designed to maximize parameter observability in different degrees, varying their frequency content and excitation characteristics.
• Figure 2: Simulation and estimation pipeline overview. (Top left) Example of satellite dynamics and actuation under a sinusoidal torque profile: body-frame angular velocity, reaction wheel speeds, commanded vs. actual applied torques, and total/split kinetic energy evolution. (Top right) Axis-wise EKF inertia tracking against ground truth for a time-varying inertia scenario, along with final absolute estimation errors compared to the LS method. (Bottom) EKF tracking performance for angular velocity, estimated inertia components, and reaction wheel speeds over time. Together, these plots illustrate (i) the excitation characteristics and physical variables simulated, (ii) the EKF’s capability to track both kinematic states and inertia parameters, and (iii) the relative estimation accuracy across methods.
• Figure 3: Mean and std deviation of inertia estimation for static inertia scenario, grouped by satellite configuration and excitation profile. Error bars represent variability across seeds, highlighting conditions where each estimator performs more reliably.
• Figure 4: Mean and standard deviation of inertia estimation errors for dynamic scenarios across satellite configurations (columns) and dynamic inertia modality (rows). Results are averaged over 10 seeds, with error bars showing variability.