A Meta-modeling Approach to Failure Prognosis Using Existing Data

Impact Technologies, LLC, proposes to develop and demonstrate an A Meta-modeling Approach to Failure Prognosis Using Existing Data. The approach will serve as the basis for a prognostic system framework using cause-effect meta-model Response Surface Method (RSM) concepts to capture the relationship between the prognostic outcome (RUL or TTF), and its primary factors as manifested from existing processed data. Historically, implementation of CBM/PHM algorithms for fault diagnosis and prognosis require the availability of “ground truth” fault and fault progression data derived from seeded fault testing or on-platform data acquisition. Such data are not readily available for most complex systems and fault data acquisition methods require extensive resources to arrive at an appropriate dataset. Introduction of new techniques such as the one proposed, that utilize archived data in the absence of on-line fault/failure time series data, in combination with CBM/PHM technologies and classical reliability tools, are needed to accurately predict the remaining life of military and industrial assets. In doing so, these techniques will make possible more wide spread, cost effective implementation of on-line, real-time CBM/PHM systems; thereby, greatly reducing the cost and burden associated with maintenance, repair and overhaul practices for aircraft, ground facilities, and other vital equipment. BENEFIT: Implementation of CBM/PHM algorithms for fault diagnosis and prognosis require “ground truth” fault and fault progression data typically derived from seeded fault testing or on-platform data acquisition. Such data are not readily available for most complex systems and fault data acquisition methods require extensive resources to arrive at an appropriate dataset. The Meta-modeling Approach to Prognosis using Existing Data offers a crucial link to the implementation of CBM/PHM strategies where currently not possible due to the absence of ground truth. In this approach, existing data serves as the basis for a prognostic model, circumventing more typical lengthy development paths. In doing so, this enables quicker, more cost effective development of CBM/PHM technologies and contribution to the transformation away from traditional scheduled/corrective maintenance practices to predictive maintenance. This approach also serves as the basis for a comprehensive Life Management Framework that includes: data processing/mining, FMECA/RCM methods, data/feature fusion techniques, metamodel design/testing, decision support, performance metrics, and risk assessment/management. Given the potential benefits, this approach to usage based prognostics will be attractive to military and commercial applications, including, but not limited to ground and air vehicle maintenance/repair organizations, aircraft logistics support centers such as Warner Robins, manufacturing, and ship construction/repair facilities.