Design of artificial neural networks for rotor dynamics analysis of rotating machine systems


Taplak H., Uzmay I., Yildirim S.

JOURNAL OF SCIENTIFIC & INDUSTRIAL RESEARCH, vol.64, no.6, pp.411-419, 2005 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 64 Issue: 6
  • Publication Date: 2005
  • Journal Name: JOURNAL OF SCIENTIFIC & INDUSTRIAL RESEARCH
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Page Numbers: pp.411-419
  • Erciyes University Affiliated: Yes

Abstract

A neural network predictor is designed for analyzing vibration parameters of the rotating system. The vibration parameters (amplitude, velocity, acceleration in vertical direction) are measured at the bearing points. The system's vibration and noise are analyzed with and without load. The designed neural predictor has three (input, hidden, Output) layers. In the hidden layer, 10 neurons are used for approximation. The results show that the network is useful as ail analyzer of such systems in experimental applications. The neural networks are validated for reduced test data with unknown faults.