AUTOMATIC FAULT DETECTION AND FAILURE PREDICTION FOR SPACECRAFT SYSTEMS

TIDG PROPOSES TO DEVELOPE AN INTEGRATED FAULT DETECTION AND FAILURE PREDICTION SYSTEM FOR SPACECRAFT MECHANICAL, ELECTRICAL AND STRUCTURAL SYSTEMS. OUR APPROACH USES OUR PROVEN PATTERN RECOGNITION FAULT DETECTION TECHNIQUES THAT WE DEVELOPED FOR NAVY HELICOPTER GEARBOXES COUPLED WITH NEW RESEARCH IN FAILURE PREDICTION AND MACHINE LEARNING. AN ADVANTAGE OF OUR APPROACH IS THAT PATTERN RECOGNITION DOES NOT REQUIRE A TIME HISTORY DATA BASE AND THUS ONBOARD MEMORYREQUIREMENTS CAN BE MINIMIZED. THE APPROACH ALSO HAS THE BENEFIT OF A LOW FALSE ALARM RATE. BECAUSE OF OUR PREVIOUS AND ONGOING RESEARCH IN RELATED TECHNOLOGIES, THE PROPOSED PROGRAM WILL ACHIEVE MUCH MORE THAN NORMALLY POSSIBLE WITHINTHE SBIR FUNDING LIMITS. THE OBJECTIVE WILL BE TO COMPLETE A DEMONSTRATION HARDWARE AND SOFTWARE SYSTEM BY THE END OF PHASE II.