Major Challenges in Prognostics: Study on Benchmarking Prognostics Datasets

F. Camci, O. F. Eker, and I. K. Jennions
Submission Type: 
Full Paper
Supporting Agencies (optional): 
IVHM Centre, Cranfield University, UK and its industrial partners
phmce_12_004.pdf654.52 KBMay 25, 2012 - 12:32pm

Even though prognostics has been defined to be one of the most difficult tasks in Condition Based Maintenance (CBM), many studies have reported promising results in recent years. The nature of the prognostics problem is different from diagnostics with its own challenges. There exist two major approaches to prognostics: data-driven and physics-based models. This paper aims to present the major challenges in both of these approaches by examining a number of publicly available datasets for their suitability for analysis. Data-driven methods require sufficient samples that were run until failure whereas physics-based methods need physics of failure progression. Each dataset is analyzed individually and applicability of data-driven and physics based models are discussed.

Publication Year: 
Publication Volume: 
Publication Control Number: 
Page Count: 
Submission Keywords: 
benchmarking datasets
Remaining Useful Life Estimation
Challenges in Prognostics
Submission Topic Areas: 
Data-driven methods for fault detection, diagnosis, and prognosis
Model-based methods for fault detection, diagnostics, and prognosis
Submitted by: 

follow us

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