Autonomous Fault Diagnosis and Recovery for Advanced Life Support Systems

Stottler Henke developed a diagnosis and recovery planning system for the Intelligent Systems Branch, Automation and Robotics Division at Johnson Space Center. This work focused on the deliberative elements of our health maintenance architecture for subsystems of advanced life support (ALS) systems, including physico-chemical and bioregenerative approaches to recycling water and air. Our system employs model-based reasoning and case-based reasoning to perform diagnosis and knowledge-based planning to perform recovery planning. A key feature of the ALS system domain is that lack of sensors and redundancy require crew involvement in both diagnosis and recovery. The system includes graphical knowledge editors for the entry of structural and behavioral models, cases and crew procedure plan operators. It can employ multimedia presentations to facilitate stepping crewmembers through diagnosis and recovery procedures. The model editor includes a facility for graphically describing finite state machines to create discrete models of system behavior.