Prognostics-Based Models and Reliability

Organizer and Session Chair: Aparna Huzurbazar, Los Alamos National Laboratory


Real-time Prognostics-Based Identification of Mission/Task-Capable Units in a Fleet

Nagi Gebraeel
Georgia Tech

We consider a scenario where a fleet of identical units (trucks, planes, locomotives, etc.) where each unit consists of the same critical components. The performance/degradation of each critical component is assumed to be monitored by dedicated sensors. We are interested in identifying, in real-time, the best subset (the Top-k) units that are capable of achieving a predefined mission or task based on the residual life of their critical components and subsystems. We develop a Prognostics-Based Ranking Algorithm that combines stochastic degradation models with computer science database ranking algorithms. The stochastic degradation modeling framework is used to compute and update, in real-time, residual life distributions (RLDs) of the critical components of each unit. Using a base case degradation model, we address the challenge of stochastically ordering the RLDs of similar components on the different units. Next, we utilize a database ranking algorithm, known as the Threshold Algorithm, to identify the Top-k units without necessarily computing all the RLDs. This ensures a seamless and efficient real-time implementation.


A Proactive Analytical Process for Aircraft Maintenance

I-Li Lu
The Boeing Company

In aircraft maintenance, the primary objective is to ensure realization of the inherent (design) safety and reliability levels of the aircraft. It is also intended to restore safety and reliability to the inherent levels when deterioration has occurred. In addition, an important purpose is to initiate design improvement of those items whose inherent reliability are inadequate and most importantly to accomplish these goals at a minimum cost. Boeing, in its recent endeavor and as part of the Instructions for Continued Airworthiness requirements, applying industry sanctioned maintenance steering group (MSG-3) concepts and advanced statistical reliability methods, has successfully developed and implemented a Decision Support Optimization Tool to guide engineers and airlines on determining the optimum inspection interval. In this talk, we discuss the extension of this development to Reliability Centered Maintenance (RCM) programs using metrics that are suitable for military applications. We show that there exists cost effective ways to conduct preventive maintenance. And using data from sensors, maintenance logs, and degradation measures, this RCM-based process can also be used to enables automated, condition-based maintenance.


Prognostic Models Based on Nonparametric and Semiparametric Flowgraphs

David Collins
Los Alamos National Laboratory

We present a method for developing models for predicting system health (e.g., probability of failure within a given mission duration), based on component-level reliability models. Component models may be specified as parametric probability distributions, or nonparametrically as empirical distribution functions. Flowgraph methods are then used to determine the system failure time distribution. The method is illustrated with a simple application to aircraft maintenance.