12-04, 17:00–17:20 (Europe/Luxembourg), Banquet Room
Despite decades of space exploration, the very outer regions of the solar system remain largely uncharted, leaving open questions about the interaction of heliosphere and interstellar medium. One promising interstellar precursor mission architecture which can provide the challenging delta-v needed involves a Solar Oberth Maneuver (SOM) using solar electric propulsion, followed by a Jupiter gravity assist.
Devising an effective steering strategy for such a mission is challenging. Using evolutionary neurocontrol, previous research demonstrated the feasibility of flight times under 25 years (Loeb et al, 2011); yet, with the necessity of using a Jupiter Gravity Assist (JGA) and including a further Radioisotope Electric Propulsion (REP) stage which reduces payload mass to 35 kg. However, advancements in high-temperature solar cells might allow for SOMs even closer than 0.7 AU, and the new Falcon Heavy launcher from SpaceX for higher payload capabilities.
Previous implementations of evolutionary neurocontrollers are often sophisticated, involving artificial neural networks (ANNs) with high-dimensional inputs (up to 28 parameters) and 4 output nodes. The high-dimensional input domain means that the algorithm first needs to learn which parameters to weigh more, and which not, which is suspected to elongate the optimization process.
The work at hand has two main objectives:
i) developing a streamlined evolutionary neurocontrol architecture for optimizing SOM steering strategies to achieve lower bound payload estimates and;
ii) using it to evaluate payload capabilities of a SOM to the heliopause with a perihelion distance of 0.3 AU, assuming a Falcon Heavy launch and an electric thruster with specific impulse of 6000s and 75% thrust efficiency.
A mutation-driven evolutionary algorithm is used to optimize the steering strategy given by a neurocontroller that transforms the current spacecraft state into steering control commands. It is shown that a neurocontroller with 4 input parameters, 20 hidden nodes and 2 output nodes suffices for lower bound assessments of payload capabilities to 200 AU within 25 years. It is found that the SOM at 0.3 AU allows for a mission scenario without JGA and REP stage, yielding at least 1,551 kg payload mass.
Moreover, a minimum specific power of the solar panels is derived, indicating that present-day specific solar panel masses are sufficient, given a tolerance of 400°C operational temperature during the SOM.
On the condition of sufficient maturity of the high-temperature solar cells, these results suggest feasibility of a short- or mid-term heliopause mission without the necessity for a REP stage or JGA, thus allowing for a mission of decreased complexity, higher scientific payback and more flexible launch window.
Nadim Maraqten is an aerospace engineering masters student at the University of Stuttgart and TU Delft, focusing on electric propulsion systems and machine learning.
Since 2022 he is part of the initiative for interstellar studies and conducts studies on advanced electric propulsion concepts.
Moreover, he is part of the team behind the MVSE Venus exploration mission proposal, as well as the LUMO space station concept for in-space manufacturing of solar panels from lunar regolith.
