Model-based System-level Health Management for Reusable Launch Vehicles

Abstract

The next generation of reusable launch vehicles are expected to radically reduce the cost of accessing space thus enabling a broad range of endeavors including the commercialization of space and further manned exploration of the inner solar system. A reduction in operational costs requires more sophisticated techniques for monitoring and controlling the vehicle while in flight and techniques to streamline the ground processing of the vehicle. For both tasks, it is often necessary to synthesize information obtained from multiple subsystems to detect and isolate both hard failures and degraded component performance.

Traditionally, the synthesis of information from multiple components or subsystems is performed by skilled ground control and maintenance teams. In the future, much of this task will need to be performed by more sophisticated software systems that are able to reasoning about subsystem interactions. Performing this task using a traditional software programming paradigm is challenging due to the myriad of interactions that occur between subsystems especially when one of more components are performing in some off-nominal fashion. Model-based programming addresses this limitation by encoding a high-level qualitative model of the device being monitored. Using this model, the system is able to reason generatively about the expected behavior of the system under the current operating conditions, detect off-nominal behavior and search for alternative hypotheses that are consistent with the available observations. In this paper, we will talk about the Livingstone model-based health management system. Livingstone was initially developed and demonstrated as part of the Remote Agent Experiment on Deep Space One. Future flight experiments are planned for both the X-34 and the X-37 reusable launch vehicles.