Nowadays, the space structures control domain is still a very active research field, due to the continuous development of modern space activities and the consequent evolution of mission needs and requirements. High precision pointing control for flexible spacecraft with large appendages requires special care, especially as currently satellites are often required to perform increasingly fast slew manoeuvres for Earth Observation (EO) purposes. Indeed, even mild rigid body/flexible dynamics interactions can affect and deteriorate the pointing and stability performance, if a proper control strategy is not adopted. In the last decades, several techniques have been considered to address the Control/Structure Interaction (CSI) issue, such as traditional controllers coupled with filters and frequency separation design strategies. However, standard methods still present several limits and drawbacks, such as computational tuning burdens and attitude actuation limits. In this research, an alternative approach is considered by investigating an Active Vibration Control (AVC) strategy. In detail, the foreseen impact of this research on the current state-of-the-art is twofold. On one hand, it will lead to the synthesis and experimental testing of an advanced fully-integrated attitude/vibration control system, coupled with smart Failure Detection, Identification and Recovery (FDIR) on a scaled laboratory floating platform, leading to a TRL4 level for the proposed technology. Also, a promising yet still scarcely diffused vibration suppression device will be implemented in the control loop, namely an Offset Piezoelectric Stack Actuator (OPSA). Therefore, in this project, the testing of an OPSA-based controller performance for damping out elastic vibration during an attitude manoeuvre will be, to the proponent's best knowledge, the first attempt on a laboratory floating platform available in literature.
According to the current state of the art for missions with large flexible appendages or very low structural frequencies, Control/Structure Interaction (CSI) issues are solved at GNC level by designing controllers with low control bandwidths, limiting the force/torque commands, restricting the size of the flexible elements or using filters to be coupled with standard feedback controllers, thus requiring extensive computational efforts for the tuning of the controller. This study enables to go much further in the current scenario by proposing an optimised GNC including not only an active control system for large space structures, specially focused on antenna-type appendages to improve overall system performance, but also a Failure Detection, Identification and Recovery (FDIR) scheme to enhance the autonomy of the system.
On one hand, in-orbit active control is a consistently researched field, due to the numerous advantages it can bring to the system, such as the possibility to design lighter structures and to increase the agility and stability of the satellite. In this regard, this research would improve scientific knowledge of a field long engaged by space agencies and industries, tailoring the up-to-date findings to a meaningful (not merely academic) test case. Indeed, a recently proposed type of piezoelectric device, namely Offset Piezo-Stack Actuator (OPSA), with performance superior to a piezo patch (being the associated piezoelectric d33 constant higher than the piezo patch d31 constant), studied in very recent papers with promising results [1], will be used. Therefore, a consistent improvement in the current state-of-the-art will be achieved by testing such a device in closed-loop control on a floating platform reproducing the in-orbit dynamics of the satellite, with the aim of reaching a TRL4 level for the robust GNC-OPSA-FDIR control architecture. To the proponent's best knowledge, this application would be the first test of a satellite scaled antenna prototype in a laboratory simulated microgravity condition with such a type of actuator device. Therefore, this research will take another step toward space validation of piezo-electric sensors devices for space applications, contributing to making it a proven viable technology to improve future Italian and European space missions.
On the other hand, also the development of a Deep Learning (DL) framework to detect damages in space flexible structures in space conditions is a novel concept in the aerospace field and it is likely to lead to innovative solutions for modern space activities. Indeed, this research also aims at tailoring novel DL algorithms, based on the concept of Long Short-Term Memory Neural Networks (LSTM-NN) for interfacing them with a robust controller in the framework of a FDIR architecture. Indeed, Deep Neural Network (DNN) architectures and DL techniques will be adopted in order to embed a complex and automatic feature extraction process in the training process and to make performances independent of the physical nature of data to be processed, so to be applied to both units at platform and distributed level on the flexible structures. Such a new technology could help in finding an innovative solution for a more accurate failure handling of satellite scientific instruments, as, for instance, optical payloads mounted at the end of very long masts requiring high dimensional stability that have survived a debris impact and changed their properties.
Moreover, an in-house open-source code will be created in Matlab to be interfaced with most diffused commercial simulations software, in order to perform accurate and precise analysis, also involving piezoelectric nonlinear elements. Indeed, once the architecture of the LSTM-NN has been realized in Matlab, it will be prepared in order to receive input/output data directly from a high-fidelity in-orbit simulator reproducing synthetic satellite telemetry data, so as to validate the concept on the floating platform.
[1] W. Li, Z. Yang, K. Li, W. Wang, Hybrid feedback PID-FxLMS algorithm for active vibration control of cantilever beam with piezoelectric stack actuator, Journal of Sound and Vibration, Vol. 509, 29 September 2021