Classification of pitting fault levels in a worm gearbox using vibration visualization and ANN
| Autor: | Rafet Can Ümütlü, Hasan Ozturk, Zeki Kiral, Berkan Hizarci |
|---|---|
| Jazyk: | angličtina |
| Rok vydání: | 2020 |
| Předmět: |
Worm drive
Multidisciplinary business.product_category Artificial neural network Computer science business.industry Process (computing) Tooth surface 02 engineering and technology Structural engineering Fault (power engineering) 01 natural sciences Shock (mechanics) Vibration 0103 physical sciences 0202 electrical engineering electronic engineering information engineering 020201 artificial intelligence & image processing business Reduction (mathematics) human activities 010301 acoustics |
| Popis: | Mechanical power transmission systems are an indispensable part of the industrial process. The most complex equipment of these processes is the gear systems. Among the gear systems the worm gearboxes are used in various applications, especially those that need high transmission ratios in one reduction stage. However, worm wheel manifests defects easily because it is made of soft material, in comparison with the worm. The stress on each tooth surface may increase because of overload, shock load, cyclic load change, gear misalignment, etc. This often causes pitting faults in worm gearboxes. This paper focuses on the detection of localized pitting damages in a worm gearbox by a vibration visualization method and artificial neural networks (ANNs). For this purpose, the vibration signals are converted into an image to display and detect pitting defects on the worm wheel tooth surface. In addition, statistical parameters of vibration signals in the time and frequency domains are used as an input to ANN for multi-class recognition. Later, the results obtained from ANN are compared for both axial and radial vibration. It is found that the ANN can classify with high accuracy for any sample of the vibration data obtained from the radial direction according to fault severity levels. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|