Diagnostic Models for Failure Analysis and Operations
- Created: Tuesday, 01 February 2011
The GDP combines three existing tools. The first tool is TEAMS, described above. The second tool, SHINE (Spacecraft Health Inference Engine), is a rule-based expert system that was developed at the NASA Jet Propulsion Laboratory. SHINE rules were developed for failure detection and mode identification, and SHINE outputs served as inputs to TEAMS. The third tool, IMS (Inductive Monitoring System), is an anomaly detection tool developed at NASA Ames Research Center. The GDP was deployed to KSC and monitored live data during the prelaunch period leading up to the October 28, 2009 launch of Ares I-X. Ares I-X did not have any failures in the systems monitored by the prototype. The GDP had a small number of false alarms, largely due to differences between the Ares I-X data and the historical Space Shuttle data on which IMS was trained. The prototype successfully demonstrated the feasibility of integrating three very different failure detection and fault diagnostic methods, and of integrating diagnosis of the vehicle with diagnosis of the ground systems.
Using lessons from the Ares I-X GDP, an FDIR prototype application for the Constellation liquid hydrogen subsystem was developed to provide the initial operating capability for KSC’s launch control system. It included TEAMS for diagnostics and IMS for anomaly detection. The FDIR prototype used a liquid hydrogen system diagnostics model, an Ares I Main Propulsion System diagnostic model, and a ground power system model to formulate preliminary processes and interface requirements for ground-vehicle and ground-ground diagnostic model integration and validation. The FDIR Architecture also investigated the feasibility of integrating prognostics (failure prediction) capabilities.
Benefits of Automated Diagnostics
The automation of pre-launch diagnostics for launch vehicles offers three potential benefits: improving safety, increasing launch availability, and reducing cost. In today’s launch processing environment, fault isolation is conducted on a subsystem-by-subsystem basis. Launch support personnel identify and respond to anomalies, faults, and failures by conducting complex design analyses, tracing failure effect propagation paths, and correctly identifying suspected or bad components in real time, without benefit of automation to assist with integrated analysis. Without such automation, the complexity of ground and vehicle systems and their interactions requires a large, highly skilled workforce for safe operations.
Integrated diagnostic models developed by system designers provide accurate and rapid (on the order of seconds instead of minutes, hours, or days when not automated) information on locations and identities of potential causes of the observed failure effects. Faster diagnosis decreases recovery time and increases the launch availability of ground and vehicle systems. Integrated diagnostic models also support flight rationale assessments. Encapsulating design knowledge within the diagnostic models reduces operations personnel workload and enables more efficient launch operations. Anomaly detection and prognostics applications also increase launch systems availability and decrease workforce requirements by alerting operators to anomalous conditions and impending failures. Both techniques enable condition-based maintenance, which prevents future system damage and reduces remediation time and cost.
The FDIR Architecture and diagnostic tool suite will be matured in follow-on technology development efforts to perform integrated diagnostics for ground-vehicle systems, and improve troubleshooting and recovery by providing operators with recommendations for mitigation and/or recovery from anomalous or failed conditions. Diagnostic models, in their analytic and operational uses, provide significant benefits to NASA programs, improving safety, reliability, and availability. If the models are developed in design and re-used in operations, their development costs are significantly decreased compared to current NASA methods, and they provide significant cost benefits in operations by reducing diagnostic and decision-making times, and reducing operator workloads. These analytic and operational diagnostic capabilities are expected to be certified to support the 21st Century Space Launch Complex, and are applicable to other applications beyond launch vehicles, including surface systems, crewed spacecraft, robotic spacecraft, and aircraft.