Data Qualification and Validation Testbed--New Tool Developed and Demonstrated for Evaluating the Performance of Engine Health Management Systems

Online data validation is a performance-enhancing component of modern engine control and health management systems. It is essential that the performance of a data-validation system be verified prior to its use in a flight-rated engine control and health management system. A new Data Qualification and Validation (DQV) Testbed application was developed and demonstrated. This testbed provides a systematic test environment for that performance verification. It was used to evaluate a model-based data-validation package being employed as the data-validation component of a rocket engine health management system. The DQV Testbed was developed and evaluated through an in-house effort at the NASA Glenn Research Center in collaboration with Christian Brothers University and Expert Microsystems, Inc.

The DQV Testbed is composed of four major modules: the test manager module, the data source module, the data-validation module, and the analysis and reporting module. The functionality of these modules and the information flow between them is illustrated in the diagram. The test manager module is used to define the test conditions, and it controls the overall execution of the test sequence. The data source module generates test data for the data-validation module. The data-validation module acts as an interface between the testbed and the health management system being evaluated. The analysis and reporting module evaluates the output from the system being tested with the known test conditions and generates a series of reports summarizing the results.

Functionality and information flow within the DQV Testbed.
Long description of figure.

The primary benefit of the DQV Testbed is its ability to execute numerous health management system evaluations and to present results in a concise, yet informative manner. The reporting of results can be customized for a particular health management system as in the sample report shown in the table. Alternatively, a broad-based report can provide evaluation results in a histogram format using five generic categories:

1. Correct fault detection and isolation--A fault condition was detected and isolated to the one correct fault.
2. Fault detection and partially correct isolation--A fault condition was detected and isolated to a fault candidate list that included the correct fault.
3. Fault detection, but incorrect isolation--A fault condition was detected, but it was isolated to a fault candidate list that did not include the correct fault.
4. No fault detection--A fault condition was not detected.
5. False fault detection--A fault condition was incorrectly detected when a fault was not inserted.

Results like these provide a means for developers to characterize the performance of and assess tradeoffs for a given health management system.

During initial studies, the DQV Testbed was shown to be an effective tool for cost-effective and comprehensive testing of health-monitoring systems. It provides an efficient avenue for assessing the strengths and weaknesses of prototype data-validation and fault-detection systems by improving the understanding of those systems’ capabilities and tradeoffs. The testbed provides a potentially useful and important element of the infrastructure needed to iteratively develop and flight-qualify future propulsion health-monitoring systems.

Bibliography

Sowers, T.; Santi, L.; and Bickford, R.: Performance Evaluation of a Data Validation System. AIAA-2005-4486, 2005.

Next article

Previous article

Last updated: October 10, 2006

Responsible NASA Official: Gynelle.C.Steele@nasa.gov
216-433-8258

Point of contact for NASA Glenn's Research & Technology reports: Cynthia.L.Dreibelbis@nasa.gov
216-433-2912
SGT, Inc.

Web page curator: Nancy.L.Obryan@nasa.gov
216-433-5793
Wyle Information Systems, LLC

NASA Web Privacy Policy and Important Notices