Viability under adversity: Safe self-organizing control of systems in the unknown

On June 5, 2024, the Boeing Crew Flight Test (Boe-CFT) came to a start, launching two commercial crew astronauts into space from Kennedy Space Center, at Cape Canaveral, Florida. While spirits were initially high, as the mission progressed, flight controllers became aware of the fact that the Boeing Starliner had suffered multiple helium leaks, causing five of its 28 reaction control thrusters to fail. Similar issues occurred during the Boeing Orbital Flight Test 2 (Boe-OFT 2) of May 19, 2022. While the cause of these leaks is not fully known yet, NASA and Boeing investigators acknowledged that the four helium manifold leaks are a systemic problem with the propulsion system, as opposed to an isolated issue cause by a single defective seal. Ultimately, the Starliner successfully docked to the Harmony module of the International Space Station (ISS) after a series of hot-fire tests concluded in four out of the five defective thrusters becoming operational again. The thruster issues resulted in over an hour in delays, as well as a period of manual pilotage during which the crew performed stationkeeping outside of the 200-meter keep-out zone of the ISS. This leads us to question what would have happened if these thrusters did not come back online, and if the impairment to the spacecraft was more complex. Our goal in this work is therefore to develop a framework that enables systems to survive, or be viable, in the face of adversity. We study the effects of impairments in control authority and sensor outputs, as well as wholesale changes in the dynamics on the performance and safety of a system, in the case that the exact impairment is not fully characterized. Throughout this work, we are interested in guaranteed properties: ones that we can certify with limited knowledge about the form of degradation, all the while the system is in operation. We first quantify the impaired system's behavior based on its nominal behavior and rudimentary knowledge of a family of degradation modes that the system is experiencing, through guaranteed reachability analysis. We proceed by extending these results to long-term reachability by studying guaranteed monotonicity, which allows us closely predict the long term effects of impairment, thereby extending the system's remaining useful lifetime. We then turn to quantifying and characterizing impairment modes, with the goal of passively sharpening our knowledge of the family of degradation modes to more accurately devise mitigation strategies and predict the long-term behavior of the impaired system. Finally, we apply this acquired knowledge about the impairment modes and the long-term behavior of the system to synthesize safe controllers that enable the system to remain viable, despite the persistent adversity it may experience. We illustrate our theory on wide variety of applications, often turning back to the problem of attitude control of a spacecraft with rogue reaction control thrusters, showing how our theory can safely detect, identify, and mitigate impairment in real-time.

Name thesis:degree_name

Ph.D.

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

Aerospace Engineering

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2024

Author dc:creator

• El-Kebir, Hamza

Contributors dc:contributor

• Ornik, Melkior
• Bentsman, Joseph
• Langbort, Cedric
• Choi, Changrak

• Redundancy
• External Perturbations
• Perturbation Families
• System Dynamics Changes
• Controller Synthesis Under Uncertainty
• Degradation Modes
• Identification
• Real-time Identification
• Mitigation Strategies
• Real-time Mitigation
• Fault-tolerant Control
• Control Authority Impairment
• Sensor Output Impairment
• Safety
• Performance Reduction
• Viability In Adversity
• Reachability In Impaired Systems
• Resilience In Control Systems
• Safe Controllers
• Control Under Adversity
• Guaranteed Reachability
• Monotonicity In Control Systems
• Degradation Modes Identification
• Attitude Control
• Spacecraft Control
• Aerospace Engineering
• Control Systems
• Fault Detection
• Fault Mitigation
• Safety-critical Systems
• Gradual Degradation
• Impairment
• Fault Tolerance
• Robust Control
• Adaptive Control
• Viability Theory
• Reachability Analysis
• Guaranteed Properties
• Resilience
• Self-organizing Systems

Statement dc:rights

• Copyright 2024 Hamza El-Kebir

Language dc:language

en, eng

Handle dc:identifier

https://hdl.handle.net/2142/127201

OAI record oai:identifier

https://www.ideals.illinois.edu/oai-pmh?verb=GetRecord&metadataPrefix=etdms&identifier=oai%3Awww.ideals.illinois.edu%3A2142%2F127201