SBIR/STTR Award attributes
Back-up power equipment, such as electrical transformers, undergo performance degradation over time that can prevent them from functioning properly in critical situations, or even cause catastrophic failure. Improved diagnostic and prognostic approaches that more accurately determine the reliability and performance of transformers can advise the maintenance or replacement of power systems to prevent failures in critical situations, reduce costs, and protect against future threats. Under an initial Phase II effort, we designed and developed the Probabilistic Operations Warranted for Energy Reliability Evaluation and Diagnostics (POWERED) application, which uses rich, modular probabilistic modeling to advise reliability and maintenance of transformers. POWERED uses a distributed network of continuous-time dynamic probabilistic relational models (PRMs) to accurately and efficiently diagnose and predict system- and component-level failures due to short-term and long-term degradation. POWERED’s objectives are aligned with the Army’s overall vision of energy informed operations that provide a reliable and uninterrupted power supply with improved operational efficiency. Our proposed sequential Phase II effort expands the scope of our initial Phase II which pivoted from generators to transformers, developing a full-scope prototype and validating our approach using rich data from multiple transformers.
