Towards a Methodology for Design of Prognostic Systems

Jose Ignacio Aizpurua and Victoria M. Catterson
Submission Type: 
Full Paper
AttachmentSizeTimestamp
phmc_15_022.pdf275.79 KBAugust 18, 2015 - 2:06am

An effective implementation of prognostic technology can reduce costs and increase availability of assets. As a result of the rapidly growing interest in prognostics, researchers have independently developed a number of applications for asset-specific modeling and prediction. Consequently, there is some inconsistency in the understanding of key concepts for designing prognostic systems. This further complicates the already-challenging design of new prognostic systems. In order to progress from application-specific solutions towards structured and efficient prognostic implementations, the development of a comprehensive and pragmatic methodology is essential. Prognostic algorithm selection is a key activity to achieve consistency throughout the design process. In this paper we present a design decision framework which guides the designer towards a prognostic algorithm through a cause-effect flowchart. Failure modes, application characteristics, and qualitative and quantitative metrics are used to determine an appropriate approach for the stated problem. The application of the methodology can reduce the time and effort required to develop a prognostic system, ensure that all the possible design options have been considered, and provide a means to compare different prognostic algorithms consistently. The framework has been applied to different prognostic problems within the power industry to illuminate its effectiveness. Case studies are presented to show how the framework guides designers through the choice of prognostic algorithm according to system requirements. The results demonstrate the applicability of the methodology to the design of prognostic systems which consistently meet the established requirements.

Publication Year: 
2015
Publication Volume: 
6
Publication Control Number: 
022
Page Count: 
14
Submission Keywords: 
requirements analysis
early system design
Generic prognostics methodology
model selection
Submission Topic Areas: 
Standards and methodologies
Submitted by: 
  
 
 
 

follow us

PHM Society on Facebook Follow PHM Society on Twitter PHM Society on LinkedIn PHM Society RSS News Feed