Automated Recognition of Advanced Vibration Features for Machinery Fault Classification

Abstract

Advanced condition monitoring systems use pattern recognition and automated reasoning on features extracted from sensor data to assess the current health of a component. This paper will evaluate pattern recognition techniques for classifying the "stage of fault" using transitional failure data for commercial grade gearboxes. Features will be extracted from accelerometer data obtained on the Mechanical Diagnostic Testbed (MDTB) at Penn State Applied Research Lab. The ARL CBM Features toolbox, a MATLAB-based toolbox containing most of the traditional HUMS features and several novel features, will be used to perform feature extraction. Several classifiers and training methods will be evaluated, as well as the effect of using different dimension-reduction techniques on classification. The results obtained using the transitional failure data sets will contribute to enhanced health monitoring techniques and improved machinery health prognostic estimates.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Apr 05, 2001
Accession Number
ADP013497

Entities

People

  • Amulya Garga
  • Gregory Babich
  • Jeffery Banks
  • Katherine Mcclintic
  • Robert Campbell

Organizations

  • Pennsylvania State University

Tags

Communities of Interest

  • Materials and Manufacturing Processes
  • Sensors

DTIC Thesaurus Topics

  • Artificial Intelligence
  • Artificial Intelligence Computing
  • Data Sets
  • Dimensionality Reduction
  • Feature Extraction
  • Information Science
  • Machine Learning
  • Neural Networks
  • Pattern Recognition
  • Probability
  • Probability Density Functions
  • Reliability
  • Statistical Analysis
  • Test And Evaluation
  • Test Facilities
  • Test Methods
  • Two Dimensional

Fields of Study

  • Engineering

Readers

  • Computer Vision.
  • Fault Tolerant Diagnosis of Black and White Balloon Isolation Tests Using ¥.

Technology Areas

  • AI & ML