Force analysis of bearings on a modified mechanism using proposed recurrent hybrid neural networks

YILDIRIM Ş. , ESKİ I. , Kalkat M.

JOURNAL OF MECHANICAL SCIENCE AND TECHNOLOGY, vol.22, no.7, pp.1323-1329, 2008 (Journal Indexed in SCI) identifier identifier

  • Publication Type: Article / Article
  • Volume: 22 Issue: 7
  • Publication Date: 2008
  • Doi Number: 10.1007/s12206-008-0415-8
  • Page Numbers: pp.1323-1329


Due to different load conditions on four-bar mechanisms, it is necessary to analyze force distribution on the bearing systems of mechanisms. A proposed neural network was developed and designed to analyze force distribution on the bearings of a four bar mechanism. The proposed neural network has three layers: input layer, output layer and hidden layer. The hidden layer consists of a recurrent structure to keep dynamic memory for later use. The mechanism is an extended version of a four-bar mechanism. Two elements, spring and viscous, are employed to overcome big force problem on the bearings of the mechanism. The results of the proposed neural network give superior performance for analyzing the forces on the bearings of the four-bar mechanism undergoing big forces and high repetitive motion tracking. This continuation of simulation analysis of bearings should be a benefit to bearing designers and researchers of such mechanisms.