A Semi-Supervised Feature Selection Approach for Fault Diagnostics in Evolving Environments

Yang Hu, Piero Baraldi, Francesco Di Maio, and Enrico Zio
Submission Type: 
Full Paper
AttachmentSizeTimestamp
phmec_16_042.pdf615.87 KBJune 28, 2016 - 12:27am

This paper introduces a Semi-Supervised Feature Selection (SSFS) approach for selecting the most suitable features for fault diagnostics in evolving environments. The effectiveness of the proposed SSFS approach is verified with respect to an application concerning the classification of the defect type of bearings in Fully Electric Vehicles operating at different loads. The results show that SSFS allows adapting the diagnostic model to the varying load by updating the set of features used for the classification and achieves more satisfactory diagnostic accuracy than the traditional diagnostic models. The proposed diagnostic approach can contribute significantly to the maintenance practice of components such as gearboxes, alternators, shafts and pumps, whose working conditions are usually characterized by evolving environment.

Publication Year: 
2016
Publication Volume: 
7
Publication Control Number: 
042
Page Count: 
8
Submission Keywords: 
feature selection; fault diagnostics; evolving environment
Submission Topic Areas: 
Data-driven methods for fault detection, diagnosis, and prognosis
Submitted by: 
  
 
 
 

follow us

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