Electronic Prognostics - A Case Study using Switched-Mode Power Supplies (SMPS)

Increased awareness of potential cost savings and improved reliability associated with condition based maintenance (CBM) for avionic systems has generated interest in the research and development of novel electronic prognostic and health management (PHM) solutions. This paper describes the process, with related examples, used to develop prognostics algorithms for a commercially available switched-mode power supply (SMPS) using corroborative evidence sources. The process begins with a Pareto analysis indicating the primary modes of failure. Critical components are identified using a three-tier failure mode and effects analysis (FMEA) by investigating device, circuit, and system parameters sensitive to degradation. Once acceleration factors, or sources of degradation, are known damage accumulation failure models for each critical component are derived from highly accelerated life tests (HALT). Then, healthy components are systematically degraded to varying levels of severity by performing highly accelerated stress testing (HAST). These components are used in seeded fault tests to identify system- level parameters sensitive to device damage. Features extracted from data recorded during seeded fault tests are used to derive feature- based failure models. Finally, reasoning and data fusion algorithms are applied to both models to generate corroborative remaining useful life (RUL) predictions.