Improving the Reliability of Spacecraft Attitude Control Systems through Dynamic Failure Management

Characterizing and improving spacecraft reliability is essential for future deep-space missions, especially as these vehicles become more autonomous and rely on their own failure and health management systems to make decisions critical to mission success. One of the most common methods for modeling the reliability of a complex, redundant system, such as a spacecraft Attitude Control System (ACS), is through the use of a Dynamic Fault Tree (DFT), which allows for the incorporation of not only redundant system resources, but also the dynamic nature of certain failure modes. An important aspect of performing this overall system reliability analysis using a DFT is the ability to monitor the system in real time and make dynamic updates to the failure probabilities, capacity degradation profiles, and Remaining Useful Life (RUL) predictions for each system component. This notion of a partially degraded component or subsystem is important for two reasons: (1) it provides insight into the system performance and reliability, and (2) it provides a mechanism for the failure management system to use that data and predict when the component will fail. This research addresses three fundamental gaps in the current body of knowledge of applying DFTs to complex systems, in which components may degrade or experience partial failure, but are still able to perform their tasks. The first of these gaps is the ability to model partially degraded components within the DFT itself, in terms of both available capacity as well as failure Probability Density Functions (PDFs) and Cumulative Distribution Functions (CDFs). The second research gap addresses the ability of the overall control system not only to be able to dynamically calculate the capacity degradation profiles and failure PDF and CDF profiles, but also to integrate them directly with a DFT so that the system reliability can be determined dynamically and real-time optimization decisions can be made by the control system. And finally, although DFTs have been in existence for quite some time, their usage in any spacecraft subsystem has been limited, and have not been applied to any long-duration space missions. In order to address these fundamental gaps, two new DFT gates have been introduced into the overall fault tree infrastructure: the Capacity Gate and the Preference Gate. This represents a novel approach to improving system reliability by allowing a system to optimize redundant resources that may have partially degraded capacity, but that can continue to function in their intended environment. By utilizing dynamically updated capacity profiles, RUL estimations, and failure PDFs and CDFs directly within the DFT, these new gates allow the system to make real-time decisions based on the currently available capacity of system resources. The addition of these new DFT Capacity and Preference gates to the fault tree infrastructure described in this research, coupled with their integration into the overall control system and optimization approach, leads to a marked increase in both system reliability and RUL. This results in an extended mission duration, saving both time and money in deep space exploration, as well as increasing the potential for new scientific discovery.