Aplicación de técnicas de Machine Learning para la predicción de posibles averías de correas en equipos rotatorios

  1. Camacho Espino, Jorge 1
  2. Marichal Plasencia, G Nicolás 1
  3. Ávila Prats, Deivis 1
  4. Hernández López, Ángela 1
  1. 1 Universidad de La Laguna
    info

    Universidad de La Laguna

    San Cristobal de La Laguna, España

    ROR https://ror.org/01r9z8p25

Proceedings:
Congreso Iberoamericano de Ingeniería Mecánica-CIBIM 2022

Year of publication: 2022

Type: Conference paper

DOI: 10.5944/BICIM2022.019 GOOGLE SCHOLAR lock_opene-spacio editor

Abstract

In this work, Artificial Intelligence techniques are used to obtain a classification of the state of belts wear, motivated mainly by the operation of this equipment in the high pressure pumps of reverse osmosis desalination plants. Data was taken from 30 belts with different degrees of wear. A laboratory scale equipment of a high-speed rotary machine based on modifications of a column drill was utilized to test. The belts study is based on vibration measurements, which were recorded with a triaxial piezoelectric accelerometer and a dynamic signal analyzer. These signals are subjected to a preprocessing based on the analysis of the frequency domain, in order to then apply Machine Learning techniques and obtain a classification of the state of belt wear. The results show that it is possible to predict if the belt is still able to perform its functions correctly or if a substitution is necessary.