CALCE begins draft of PHM Roadmap
Prof. Pecht presented a draft roadmap for PHM at the PHM Consortium meetings on October 16–17. The consortium members were excited at the direction that CALCE was taking with regard to the research and
development of PHM technology and at the potential of added value to their organizations. The draft of the roadmap includes:
• Develop software for analysis of sensor data for probabilistic physics-of-failure PHM analysis.
• Develop and verify a complete foundation for PoF-based PHM analysis of electronics, from silicon to systems. While the intention is to keep the coverage broad, the focus will be on key failure mechanisms and PoF models.
• Aid in the development of integrated PHM sensor systems.
• Develop canaries for key failure mechanisms in electronics.
• Develop and verify a complete suite of tools for data-driven methods that can be implemented on electronic systems. Data-driven algorithms that are being currently used include Mahalanobis distance,
Euclidean Distance Clustering, Principle Component Analysis, Support Vector Machine, Baysian Support Vector Machine, Singular Value
Decomposition, Multivariable State Estimation Technique, Symbolic Time Series Analysis, Self-Organizing Map, and Fuzzy C-Means (Fuzzy Clustering).
• Develop and verify fusion prognostics methodology.
• Develop and verify ROI and cost modeling software for PHM. Future versions of the tool will incorporate calculation of ROI relative to things other than unscheduled maintenance; will include a cost penalty structure; will incorporate design for availability modeling including an enabler for a linking function to maintenance activity and for modeling fault-tolerant LRUs and systems; and will provide an option for incorporating redundancies.
• Develop PHM methodology to solve No Fault Found (NFF) problems with a focus on intermittent failures that can lead to NFF.
• Develop PHM methodology for effective product qualification.
• Aid companies in PHM implementation.