Neural networks for the monitoring of rotating machinery
- Tennessee Univ., Knoxville, TN (United States). Dept. of Nuclear Engineering
Vibration monitoring of components in engineering systems and plants involves the collection of vibration data and detailed analysis to detect features which reflect the operational state of the machinery. The analysis leads to the identification of potential failures and their causes, and makes it possible to perform efficient preventive maintenance. This paper describes a methodology for the automation of some of the activities related to motion and vibration monitoring in these systems. The technique involves training a neural network to model the inter- relationship between signals from two related sensors mounted on an engineering system or component at a time when it is known to be operating properly. Then one signal (or its characteristics) is put into the neural network model to predict the second signal (or its characteristics). This predicted signal is continuously compared with the actual signal A deviation between the predicted and actual signal indicates a changing relationship, usually failure of the component or system. This deviation may be quantified and provides meaningful information about the degree of degradation and deterioration of the component.
- Research Organization:
- Tennessee Univ., Knoxville, TN (United States). Dept. of Nuclear Engineering; Oak Ridge National Lab., TN (United States)
- Sponsoring Organization:
- USDOE, Washington, DC (United States)
- DOE Contract Number:
- FG07-88ER12824; AC05-84OR21400
- OSTI ID:
- 10112097
- Report Number(s):
- CONF-920538-28; ON: DE93003549
- Resource Relation:
- Conference: 8. power plant dynamics, control and testing symposium,Knoxville, TN (United States),27-29 May 1992; Other Information: PBD: [1991]
- Country of Publication:
- United States
- Language:
- English
Similar Records
Vibration monitoring of EDF rotating machinery using artificial neural networks
Vibration monitoring of EDF rotating machinery using artificial neural networks