Detección de Fallos de Máquinas Rotativas Mediante Histogramas de Señales Acústicas

Translated title of the contribution: Fault Detection of Rotating Machines Using Histograms of Acoustic Signals

Research output: Contribution to conferencePaper

Abstract

The rotating machines suffer wear of their components due to their normaloperation, therefore, with the passing of time they are exposed to failures. The earlyidentification or prediction of such failures complements the process of continuousimprovement of productivity. As a maintenance technique, a method of comparinghistograms of acoustic signals is proposed, which makes it possible to discriminatebetween a normal operating state and a faulty one. The acoustic signals of normal stateare stored as support vectors, with the aim of generating a normal condition signature as astandard histogram profile. The histograms of the new signals to be evaluated arecompared with the pattern by a similarity factor based on the area under the curve of theintersection of the histograms. Dissimilarity is finally proposed as an indicator of machinefailure. The evaluation of the method results in a performance of 9.2% similarity for theexperiment with a severe failure of the machine, for the experiment under normalconditions a performance of 20.2% similarity. These results indicate some variation thatallows determining failures in bearings and gears in different speed conditions.
Translated title of the contributionFault Detection of Rotating Machines Using Histograms of Acoustic Signals
Original languageSpanish (Ecuador)
StatePublished - 26 Oct 2018
EventIX Congreso Latinoamericano de Ingeniería Mecánica (COLIM 2018) - CO
Duration: 28 Nov 201830 Nov 2018
https://www.unipamplona.edu.co/colim2018/

Conference

ConferenceIX Congreso Latinoamericano de Ingeniería Mecánica (COLIM 2018)
Period28/11/1830/11/18
Internet address

Keywords

  • Acoustic signal.
  • Intersection area
  • Support vector
  • Vibration

CACES Knowledge Areas

  • 827A Industrial maintenance

Fingerprint

Dive into the research topics of 'Fault Detection of Rotating Machines Using Histograms of Acoustic Signals'. Together they form a unique fingerprint.

Cite this