A VIBRATION BASED FAULT DETECTION OF GEAR SYSTEMS USING NEURAL PREDICTOR


ULUS Ş., SUVEREN M., ERKAYA S.

International Conference on Advances in Mechanical Engineering ICAME 2016, İstanbul, Turkey, 10 - 13 May 2016, pp.426-431

  • Publication Type: Conference Paper / Full Text
  • City: İstanbul
  • Country: Turkey
  • Page Numbers: pp.426-431
  • Erciyes University Affiliated: Yes

Abstract

Condition monitoring of mechanical systems are very important for increasing the safety, reducing the high cost and the stoppage time of production, eliminating the possible fault. In this study, a gearbox system is used to investigate the vibration characteristics of possible faults. An experimental test rig is designed for measuring the system vibration under different working conditions. Some faults frequently encountered in gearbox systems are implemented. Faulty conditions are observed for the experiments. A neural predictor was designed to define the possible sources of faults in the systems. Vibration characteristics of faults are used for training,testing and validation of network weights. The results show that, neural predictor can clearly give the information about the fault types.