2025-07-24 –, Main Room 1 (Main stage)
The Brazilian National Institute for Space Research (INPE) is developing a new satellite, Amazonia-1B, with a different payload than it was used on Amazonia-1. This new configuration requires new control gains and parameters. A new control gain tuning tool, developed using Julia, enabled faster and more efficient gain selection than the previous approach. It utilizes libraries from the Julia ecosystem and a simplified single-axis model, significantly reducing the time for gain optimization.
The Brazilian National Institute for Space Research (INPE) is currently developing a new satellite named Amazonia-1B. The spacecraft bus is nearly identical to the previous mission, Amazonia-1, which was successfully launched in February 2021. However, the payload for Amazonia-1B is significantly different from that of its predecessor.
The attitude control subsystem (ACS) of a satellite is responsible for pointing the satellite towards the desired location to acquire the mission data. This subsystem is tasked with maintaining the pointing accuracy sufficiently high to ensure that the acquired data quality meets the mission requirements. In the case of Amazonia-1B, the payload is a camera with a resolution of 20 m. Therefore, the ACS must be capable of maintaining the attitude within 0.0005° during the camera integration time to prevent image blurring.
Among the numerous factors that must be accurately calibrated to maintain the necessary stable attitude, one of the most crucial is the precise adjustment of the control gains and parameters. The Amazonia-1 and Amazonia-1B missions employ a proportional-derivative (PD) controller, along with a set of filters, to mitigate disturbances. However, due to the distinct inertia characteristics of the two missions, it is not feasible to utilize the previously designed values.
In the preceding mission, MATLAB was employed to design control gains utilizing consolidated tools. However, our approach proved to be excessively time-consuming for this scenario. The tools and theoretical framework assume a linear system composed of a series of transfer functions. Nevertheless, the actual satellite incorporates numerous impactful non-linearities that substantially alter the system time response. For instance, during the design phase, the available tools did not account for actuator limitations, and the selected reaction wheel only provides 75 mNm of torque. Consequently, the set of gains designed resulted in significantly different time responses when tested in the actual hardware with a highly detailed simulation. To address this scenario, we necessitated selecting a substantial number of candidates for the gains (more than 20) and selecting the best ones based on real-time simulation, which was extremely time consuming.
For our current mission, we have adopted a more advanced approach. Utilizing Julia’s ecosystem, we constructed a control gain tuning tool employing SatelliteToolbox.jl, ReferenceFrameRotations.jl, ControlSystems.jl, ModellingToolkit.jl, and GLMakie.jl. This tool encompasses both conventional figures of merit for assessing control stability and robustness (gain and phase margins) and a time response derived from a highly precise model. The latter was obtained by simplifying our validated simulator and restricting the dynamics to a single axis. This tool allowed an effortless selection of a set of control gains that meet the specified margins while simultaneously achieving an exceptionally good time response. In this scenario, Julia’s performance proved highly advantageous, as it enables the instantaneous update of the time simulation upon the selection of a new gain set.
This tool enables us to obtain the control gains and filter parameters for the four satellite control modes significantly faster than the previous approach. The initial set of parameters obtained yields an excellent output in the real-time simulation, substantially reducing the time required to complete this task.
This presentation will provide a comprehensive description of the development process. We will demonstrate the tool and elucidate our methodology for designing satellite control gains. Additionally, we will demonstrate how we integrated the ecosystem packages to create this application. This information is highly valuable for various engineering endeavors beyond the space domain, as it can be readily adapted for tuning control gains in other types of scenarios.
