This paper discusses data analysis approaches for effective reliability prediction using prognostics and health management methodology, which integrates sensory data with models based on either healthy training data or physics-of-failure knowledge. This approach enables in-situ assessment of a product’s performance degradation or deviation from an expected normal operating condition (i.e., the system’s “health”). It also estimates future system reliability based on systems current health state by considering its historical performances as well as operating conditions. Various studies have been conducted, to implement prognostics for electronics are summarized in this paper. These studies consider all types of electronics, ranging from components and boards to electronic products and systems.

A hybrid approach based on feature extraction and trending with physics-of-failure for prognostics is given here. This paper then discusses feature extraction, data trending and data mapping technologies that are being used by CALCE for prognostics.

Index Terms: Prognostics, Health Management, Physics of failure, Data driven, Life cycle

Complete article is available to CALCE PHM Consortium Members.