Application of Prognostic Health Management in Digital Electronic Systemsdtic.mil[PDF]

发布时间：2010-07-30 15:31:10

Abstract—Development of robust prognostics for digital electronic system health management will improve device reliability and maintainability for many industries with products ranging from enterprise network servers to military aircraft. Techniques from a variety of disciplines is required to develop an effective, robust, and technically sound health management system for digital electronics. The presented technical approach integrates collaborative diagnostic and prognostic techniques from engineering disciplines including statistical reliability, damage accumulation modeling, physics of failure modeling, signal processing and feature extraction, and automated reasoning algorithms. These advanced prognostic/diagnostic algorithms utilize intelligent data fusion architectures to optimally combine sensor data with probabilistic component models to achieve the best decisions on the overall health of digital components and systems. A comprehensive component prognostic capability can be achieved by utilizing a combination of health monitoring data and model-based estimates used when no diagnostic indicators are present. Both board and component level minimally-invasive and purely internal data acquisition methods will be paired with model-based assessments to demonstrate this approach to digital component health state awareness. 1Index Terms— Automated reasoning algorithms, physics of failure modeling, prognostic and health management (PHM)A CRONYMSAF – Acceleration FactorsBIT – Built-in TestCOTS – Commercial off-the-shelfHASS – Highly Accelerated Lift TestingMOSFET – Metal-oxide-semiconductor Field EffectTransistorMTTF – Mean Time to FailureµP – MicroprocessorPHM – Prognostics and Health ManagementPoF – Physics-of-failureRISC – Reduced Instruction Set ComputerRUL – Remaining Useful Life1 1-4244-0525-4/07/$20.00 ©2007 IEEE.Paper 1326 Version 31.I NTRODUCTIONIGITAL electronic boards are found in numerous facetsof modern day life where consumers have come todepend on their reliability to operate effectively in both both professional and private endeavors. Furthermore, the commercial and military markets demand even greater reliability constraints on semiconductor manufactures where a system failure could produce catastrophic results. Diagnostic methods have been implemented in a variety of existing electronic systems (e.g. BIT), which are effective in identifying sources of malfunctions post-failure within the system; however, fail to track system usage throughout the systems’ lifespan necessary when attempting to offer instantaneous health state assessments. A clear opportunity and vital need exists to improve digital electronic system health state awareness and prediction through development of PHM techniques.The goal of proactive fault monitoring is to prevent the end user from experiencing the effects of the failure and ideally provide advanced notice of impending failure in due time to allow corrective measures to be taken prior to failure (i.e. reduce duty cycle, offload utilization, or schedule repair). Achieving this objective requires knowledge of how component-level failure manifests throughout the system and insight as to which measurands offer indication of incipient signs of failure. In this paper, the authors illustrate how cradle-to-grave health state awareness can be achieved through the teaming of model-based assessments in the absence of fault indications and a data driven approach used to track indicators of failure providing failure mode classification. Test results from accelerated testing of a CMOS device are presented as a basis to indicate the ability to capture fault indicators indicating impending failure and track the degradation of performance measurands. The application of complementary prognostic techniques such as physics-based component damage accumulation/aging models based on projected operating conditions, empirical (trending) models, and system level failure progression models are discussed as providing a solid foundation on which to developApplication of Prognostic Health Management in Digital Electronic SystemsPatrick W. Kalgren, Mark Baybutt, Thomas DabneyAntonio Ginart, Chris Minnella, & Joint Strike Fighter Program Office Michael J. Roemer Suite 600, Crystal Gateway Four Impact Technologies, LLC 1213 Jefferson Davis Highway200 Canal View Blvd. Arlington, VA 22202-4316Rochester, NY 14623 (703) 601-5619(585) 424-1990 tom.dabney@jsf.mil patrick.kalgren@impact-tek.comD1

Application of Prognostic Health Management in Digital Electronic Systemsdtic.mil[PDF]

推荐内容

相关推荐